Michael Dell postte vorige week dat er soevereine AI-infrastructuur wordt gebouwd "op Nederlandse bodem". Het klonk alsof het lintje al was doorgeknipt. De werkelijkheid is voorzichtiger, en interessanter. VOLT opent in oktober 2026 een Nederlandse AI-cloud bij een NorthC-datacenter in Amsterdam, met apparatuur van Dell. Dat moet de eerste stap zijn naar iets veel groters: een AI-gigafabriek in de Rotterdamse haven, met plek voor zo'n 250.000 GPU's en enkele honderden megawatts aan capaciteit. ## Waarom dit ertoe doet Een gewoon datacenter host clouddiensten, databases, netwerken, allerlei bedrijfsapplicaties. Een AI-fabriek is iets anders. De racks trekken tot honderd kilowatt of meer, tegen vijf tot vijftien in een gangbare serverhal. Dat vraagt om andere koeling, andere stroomvoorziening, andere netwerkarchitectuur. Het is gebouwd voor precies het soort dichte, onderling verbonden rekenkracht die nodig is om grote modellen te trainen, niet alleen om ze te draaien. Dat verschil is waarom dit meer is dan een extra serverhal. Zonder die trainingscapaciteit blijft Europa vooral klant van andermans intelligentie: modellen gemaakt in de VS of China, gedraaid op servers die toevallig in Europa staan. Mét die capaciteit ontstaat op zijn minst de mogelijkheid om zelf modellen te bouwen, niet alleen om ze te huren. ## De schaal Als die 800 megawatt er komt, is Rotterdam een van de grootste AI-campussen van Europa. Eén Rotterdamse fabriek maakt Europa daarmee niet onafhankelijk. Maar iedere serieuze uitbreiding van de Europese rekencapaciteit verkleint de achterstand ten opzichte van de VS en China. Dat is winst, ook zonder dat het een doorbraak is. ## Wie beslist er straks VOLT is een privaat Nederlands bedrijf, opgericht door Han de Groot. Zijn family office financiert de eerste fase. Dell levert de techniek, NorthC de locatie. Over de uiteindelijke eigendoms- en financieringsstructuur van de Rotterdamse fabriek is nog niets bekendgemaakt. Dat maakt "Nederlandse bodem" een andere claim dan "Nederlandse zeggenschap", al is dat geen alles-of-niets-verschil. Een Nederlandse onderneming kan best substantiële zeggenschap houden, ook met Dell-servers en Amerikaanse chips. Werkelijke zeggenschap zit niet alleen in het eigendom van het gebouw, maar ook in financiering, technologie, contracten en toegang tot cruciale onderdelen. NorthC zelf is de afgelopen jaren al van eigenaar gewisseld tussen grote infrastructuurinvesteerders, wat laat zien dat een gebouw op dezelfde plek kan blijven staan terwijl de zeggenschap erover van eigenaar wisselt. ## Ook jouw leverancier Diezelfde gelaagdheid geldt een stap lager, bij de tools die je als ondernemer dagelijks gebruikt. Een AI-abonnement dat claimt "EU-hosting" of "data blijft in Europa" beantwoordt maar één laag van dezelfde vraag die bij Rotterdam speelt. Waar iets staat is niet hetzelfde als wie erover beslist, wie het onderhoudt, en onder welke voorwaarden. Een Europese vestiging van een Amerikaans platform kan nog altijd te maken krijgen met Amerikaans beleid, Amerikaanse prijszetting of een Amerikaans hoofdkantoor dat de voorwaarden bepaalt. Dat is geen reden om die tools te mijden, veel ervan zijn nu simpelweg de beste optie. Het is wel een reden om de claim op waarde te schatten in plaats van op geloof aan te nemen. De wezenlijke vraag is daarom niet alleen waar zo'n fabriek staat, maar wie de beslissingen neemt die uiteindelijk onze toegang, prijs en voorwaarden bepalen. Diezelfde vraag mag je gerust aan je eigen leveranciers stellen, ook als hun logo Nederlands oogt. Nederlandse bodem is waardevol. Het is alleen maar één laag van soevereiniteit. > **Let op:** Nederland kan buitenlandse investeringen in vitale en gevoelige technologie toetsen. Vanaf 2027 wordt die toets uitgebreid naar bepaalde AI-technologieën. Of een toekomstige verkoop van een AI-infrastructuurbedrijf daaronder valt, hangt af van de juridische structuur en activiteiten. Meer lezen: * EuroStack: een Europees alternatief voor je digitale gereedschap? * WhatsApp als politiek terrein * Van gratis naar betaald: hoe kies je het juiste AI-model voor je organisatie

I bought an international train ticket last week and added it to the railway app. The process took about thirty seconds. Nothing asked for my name. Nothing verified who I was. The ticket sat there, ready for a journey I haven't made yet, and I'm reasonably confident it will work fine when I hold the phone to the scanner at the platform. That small moment bothered me in a productive way. The app could just as easily have been a wallet. Buy a concert ticket and it lands in Apple Wallet or Google Wallet the same way, no identity attached. The container changes. The principle doesn't. ## Two models There are two fundamentally different ways a system can decide whether to let you in. The first is **account-based**. You are registered. The system knows who you are. Access is granted because your identity has been verified and associated with a right. A transit card linked to your name works this way. So does your bank account, your email, your government login. The credential belongs to you specifically. Lose it, and it can be blocked. Transfer it, and the transfer can be traced or refused. The second is **possession-based**. The system doesn't know who you are and doesn't need to. Access is granted because you hold something. A physical key. A fifty euro note. A paper ticket with a punched hole. The right travels with the object, not with the person. The word for this second model is bearer. Whoever bears the token gets in. 🎫 ****Bearer**** is a legal and technical term for an instrument that grants rights to whoever holds it, without verifying identity. A banknote is a bearer instrument. So is an API key. The opposite is a registered or identity-bound instrument, where the right belongs to a named holder. ## The default that was For most of human history, possession-based access was simply the only option. Institutions didn't have the infrastructure to track individuals at scale. Cash worked because it had to. The paper train ticket worked the same way. When the conductor came through the carriage and punched a hole in your ticket, they were doing two things: confirming validity and preventing reuse. The ticket changed physically. It was spent. The system was elegant precisely because it required nothing beyond the object itself. No database lookup. No identity check. No account. Just: do you have the thing? ## The drift Account-based systems became possible when infrastructure caught up. Databases, networks, identity numbers, card readers. Suddenly it became feasible to tie access to a person rather than an object. And institutions had strong reasons to prefer it. An account can be audited. A bearer token cannot. An account holder can be held responsible. A bearer token holder is anonymous by definition. The drift from possession to account wasn't a conspiracy. It was institutions doing what institutions do when the technology allows it: gaining visibility, reducing risk, building relationships with users that could be monetised or managed. Cash is the clearest example of what's at stake. It is bearer by design. The fifty euro note doesn't know who owns it. That property, which once seemed simply practical, now looks like a political position. Several European countries have introduced limits on cash transactions. The digital euro is being designed with traceability as a feature. Every move away from cash is a move from possession to account. ## Bearer didn't disappear What surprised me, once I started looking, is how thoroughly the bearer model persists inside digital systems, mostly invisible. An API key is a bearer token. Any system that receives it grants access, no questions about who is sending it. A session cookie, the small credential your browser holds after you log in, behaves the same way. Whoever holds the cookie can act as the account holder. The QR code on my train ticket is a bearer token with a one-time validation: the backend flips a flag when it's scanned, the digital equivalent of the punched hole. This is also where the distinction between human-to-machine and machine-to-machine access becomes interesting. When a script or an automated workflow authenticates to an API, it almost always does so with a bearer token. There is no biometric, no SMS code, no identity check. There is a key, and whoever holds the key gets in. The delegation problem that makes so many digital systems difficult to automate is partly a collision between these two models: systems built for account-based human identity, trying to accommodate possession-based machine access. ## The lens What changed for me after that railway app moment wasn't a conclusion. It was a way of seeing. Every time I add something to an app or a wallet, I now ask: is this bearer, or is this account? Every time I read about payment infrastructure, digital identity, or access control, the same question surfaces. The answer usually reveals something about how money and data move through the system, and who designed it to move that way. Two models. One question. It turns out to explain quite a lot.

Op donderdagmiddag stuurde het Amerikaanse ministerie van Handel een brief naar Anthropic. Enkele uren later was Fable 5, het meest geavanceerde taalmodel van het bedrijf, wereldwijd niet meer beschikbaar voor gebruikers. Niet door een storing. Niet vanwege onderhoud. Maar omdat een overheid in Washington had bepaald dat het model offline moest. De aanleiding was een zogeheten Export Control Directive: een juridisch instrument waarmee de Verenigde Staten de verspreiding van bepaalde technologie naar buitenlandse partijen kunnen beperken of stopzetten. Volgens de Amerikaanse overheid zou er een methode zijn gevonden om het model te omzeilen op manieren die veiligheidsrisico's opleveren. Anthropic betwistte die analyse en liet weten binnen 24 uur met meer informatie te komen. Op het moment van schrijven is Fable 5 nog altijd niet beschikbaar. ## Niet technisch, maar geopolitiek Voor zover bekend is dit de eerste keer dat een Amerikaans AI-bedrijf een breed gebruikt model offline heeft moeten halen op directe instructie van de federale overheid. Dat maakt dit meer dan een incident. Het legt iets bloot wat voor veel organisaties tot nu toe abstract bleef: de aard van hun afhankelijkheid. AI-tools zijn in twee jaar tijd voor veel bedrijven verschoven van handig hulpmiddel naar werkende infrastructuur. Samenvattingen, klantenservice, interne kennisbanken, geautomatiseerde workflows: steeds vaker draaien ze op modellen die ergens in een datacenter staan, aangeboden door een bedrijf dat onder Amerikaans recht valt. Dat recht geeft de Amerikaanse overheid vergaande bevoegdheden over wat bedrijven mogen aanbieden, en aan wie. Exportcontrole is een van die instrumenten. Zulke regels kunnen snel worden ingezet, vaak zonder veel aankondiging vooraf. ## Wat dit voor je bedrijf betekent Je hoeft geen paniek te zaaien om dit serieus te nemen. Maar het is wel een goed moment voor een eerlijke inventarisatie. Welke AI-tools zijn inmiddels structureel onderdeel van je werkprocessen? Wat valt er stil als een van die tools morgen niet meer beschikbaar is? Is er een alternatief? En bestaat er nog een werkbare route zonder dat specifieke model? Dat zijn geen theoretische vragen meer. ⚠️ Afhankelijkheid van AI-tools is niet principieel anders dan afhankelijkheid van SaaS of cloudinfrastructuur. Maar bij AI is de geopolitieke dimensie groter. De markt wordt gedomineerd door een klein aantal Amerikaanse bedrijven, en de beschikbaarheid van hun modellen is uiteindelijk geen zuiver commerciele kwestie. ## Een Europees alternatief? Er wordt hard gewerkt aan Europese AI-modellen. Mistral, uit Frankrijk, is de bekendste speler. Daarnaast zijn er initiatieven vanuit onder meer het European AI Office om het Europese AI-veld verder te versterken. De kwaliteit en functionaliteit van Europese modellen liggen nog niet overal op het niveau van de beste Amerikaanse modellen. Maar de kloof wordt kleiner. Voor de meeste bedrijven is een volledige overstap naar Europese modellen op dit moment waarschijnlijk nog niet realistisch. Wat wel realistisch is: bewuster omgaan met afhankelijkheid, diversifieren waar dat kan, en bij nieuwe toolkeuzes niet alleen kijken naar functionaliteit en prijs, maar ook naar zeggenschap. Wie bepaalt uiteindelijk of een tool beschikbaar blijft? Dat is geen retorische vraag meer. Afgelopen vrijdag kwam het antwoord uit Washington.

Yesterday I was using it. Today there is a message where the model used to be: "Fable 5 is temporarily unavailable." Fable 5, Currently unavailable That is how it happens. Not gradually, not with warning. At 5:21 PM Eastern on June 12, the US Commerce Department sent Anthropic a letter. By evening, Anthropic had disabled access to Fable 5 and Mythos 5 for all customers globally: the first time a leading AI company has taken a publicly deployed model offline due to intervention from the federal government. I have been on the web since the mid-nineties. The internet has always been porous, unruly, structurally resistant to central control. You could always reach something. This felt different. A switch, thrown from Washington, and the model is simply gone. ## What actually happened The US government issued an Export Control Directive, a legal order that restricts or suspends the transfer of specific products, data, or technologies to foreign countries or citizens of another country. The stated concern: a jailbreaking method for Fable 5. Anthropic reviewed the demonstration and found only a small number of previously known, minor vulnerabilities, all relatively simple, and all discoverable using other publicly available models without requiring any bypass. Anthropic called this a misunderstanding and said it would share more details within 24 hours. It is complying while disagreeing, which is the only available position when a government order arrives with legal force and no room for negotiation. To the best of available knowledge, this is the first time the United States has issued an export control directive for LLM access. The directive affects not only the US's closest allies but their nationals as well, regardless of where they happen to be. ## The personal register What surprised me was not the fact of it. I had already written, in a different context, about what happens when AI access breaks or gets cut off and listed a geopolitical rupture placing European businesses on the wrong side of US export controls as one of the non-exotic failure modes. The argument was there. The scenario was there. But knowing the argument and feeling the consequence are different experiences. I was using Fable 5 for client work. Then I wasn't. The intellectual case for sovereignty became, for a moment, a practical inconvenience. That gap, between the abstract and the operational, is where most organisations still live. ## Where it goes Whether this is temporary or precedent-setting is not yet clear. Anthropic believes it is a misunderstanding and is working to restore access. That may happen. Or it may not. What is clear is the structure underneath: the model runs in the US, the company answers to US law, and US export controls have always been able to reach foreign nationals. AI models are now, apparently, within scope. This is what I had in mind when I wrote about Lidl of all companies building its own cloud platform as a signal of where serious organisations are heading: not ideology, but a practical calculation about what you control and what you don't. The question for the rest of us is whether an event like this actually changes the calculation. Or whether, once Fable 5 is back, we return to exactly where we were.

In the summer of 1952, the first Dutch computer was switched on at the Mathematisch Centrum in Amsterdam. It filled an entire classroom. It was built from 1,200 relay switches and could store about 30 bits of information. At its launch it worked once, briefly, and was then taken out of commission. It was called the ARRA, short for Automatische Relais Rekenmachine Amsterdam. Its builders, Carel Scholten and Bram Loopstra, immediately started over. The ARRA II worked. And that same year, the Mathematisch Centrum hired a young programmer named Edsger Dijkstra. Edsger Dijkstra (left), Bram Loopstra and Ria Debets at the Mathematisch Centrum in Amsterdam (1954). ## Amsterdam Dijkstra spent the first decade of his career at the Mathematisch Centrum. In 1956, sitting in a cafe in Amsterdam without pen or paper, he worked out a solution to the **shortest path problem** in about twenty minutes. He published it three years later. It is now used in every navigation system on the planet. In 1960, together with Jaap Zonneveld, he completed the first working compiler for **ALGOL 60** , beating every other team in the world by more than a year. The Mathematisch Centrum was also, that same year, the institution from which Electrologica was spun off: the first Dutch computer manufacturer, building the X1 and later the X8 in Rijswijk. The thinking and the making were already connected, even before Dijkstra moved south. ## Eindhoven In 1962 he joined the Mathematics Department at the **Technische Hogeschool Eindhoven**. There he built an operating system for the Electrologica X8. The system was called THE, simply the initials of the university. He never chose the name deliberately. THE was built by six people, working half-time. Its architecture was organised into six strict hierarchical layers, each able to use only the functionality of the layers beneath it. Processes were coordinated using semaphores, a synchronisation primitive Dijkstra had also invented. The paper appeared in Communications of the ACM in 1968, won the ACM Programming Systems and Languages Award in 1971, and is still read in computer science courses today. The following year, Dijkstra received the Turing Award, computing's highest honour, for his contributions to structured programming. What THE demonstrated was that structure is not a constraint on ambition. It is what makes ambition hold. That logic has not aged: the systems that compound are the ones built with intention from the start. ## From Rijswijk to Veldhoven The people who built the ARRA and the X8 did not disappear when those machines did. Loopstra moved to Electrologica and later to Philips. Scholten did the same. The knowledge migrated into the industrial infrastructure of the Eindhoven region. Philips was already there. From Philips, in 1984, came ASML, starting in a leaky shed, building lithography machines. ASML now makes the only equipment capable of producing the world's most advanced chips. Every leading chipmaker depends on it. That continuity is not coincidental. It is what happens when theoretical rigour and a culture of making occupy the same small country for long enough. ## Why this is not just history Amsterdam and Eindhoven are in conversation again about working more closely on technology. The framing is **AI, talent, and infrastructure**. That framing is right. But the historical depth is worth holding on to, because it clarifies what the axis actually produces. The Netherlands has theoretical institutions of real quality. It has the most critical piece of hardware infrastructure in the global semiconductor industry. It has a tradition, visible from the ARRA to THE to ASML, of building things that matter from first principles, with limited resources, and getting them right. Europe is searching for this combination. The question of whether good ideas have somewhere to live, somewhere to scale, is unresolved. The Dutch answer to that question is already seventy years old. It is worth reading carefully.

I wrote about passkeys in early 2024 with some optimism. The technology is elegant. The problem it solves is real. I still think both those things. But I left something out. ## What passkeys actually fix The password is a structural failure. It is a secret you have to remember, type, transmit, and trust a server to store correctly. Any one of those steps can go wrong, and the weakest one determines your security. Phishing works because passwords travel. Data breaches work because servers collect them. Credential stuffing works because people reuse them. A passkey replaces all of that with a cryptographic key pair. Your device holds the private key. The service holds the public key. When you authenticate, your device signs a challenge; the server verifies the signature. Nothing is transmitted that could be intercepted or stolen. I wrote about making that transition in early 2024, and the improvement was real. It is not marketing. Back in 2022, when passkeys were still theoretical, I noted the convenience problem: helping a six- or eighty-year-old through a QR-code flow would be a challenge. What I did not see then was the deeper problem that convenience would bring with it. ## The key card stays the same Here is what does not change. Once you are through the lobby, the server hands you a session cookie: a small text file that carries your authenticated state from that point on. The elaborate verification at the front door collapses into a string of characters sitting in your browser. Copy that cookie and you are in, from anywhere, as anyone. The key card problem is untouched by how you checked in. Passkeys make the lobby better. They do not change what happens after it. ## The dependency There is a more fundamental issue. Passkeys do not live in your head, and they do not live on a neutral device. They live in a keychain: Apple's, Google's, or Microsoft's. That is where they are stored, and that is where they sync. The sync is not optional. Use more than one device and you need it. Switch from iPhone to Android and your passkeys stay behind. Add a Windows machine to an Apple setup and you are managing two separate ecosystems, each with its own rules. The line between digital convenience and infrastructural dependency is vanishingly thin -- and you have crossed it the moment you need your credentials on a second device. This is a structural dependency, not a technical inconvenience. The platform controls your credentials. If you leave, you lose them. If you are locked out, you lose them. If the platform changes its policies, you are subject to those changes. I experienced a version of this when X locked me out last year: the burden of proof shifts to you, the process is opaque, and there is no neutral party to appeal to. With passkeys, the stakes are higher. It is not one platform. It is your authentication infrastructure. ## A different kind of lock-in With passwords, the risk was theft. With passkeys, the risk is captivity. These are not equivalent problems, and the second one is quieter. Most users will not notice. The experience is genuinely better: a biometric check, no password to forget, no phishing surface. The dependency is invisible until it is not. That is how the best lock-ins work. Europe is working on a different answer, one built on the premise that digital identity should not be owned by a platform. That is a longer conversation. But it starts from exactly the right diagnosis: the front door matters, and so does who holds the key. 🔐 ****Passkeys**** are supported by Apple, Google, and Microsoft through their respective keychain and credential manager products. Cross-platform sync between ecosystems is not natively supported – moving between platforms requires manual credential migration or re-registration with each service.

A few weeks ago I ran a command that made me stop. I was downloading a video using yt-dlp, a command-line tool, and instead of logging in I passed a single flag: `--cookies-from-browser chrome`. The tool reached into my browser, borrowed my active session, and acted as me. The platform never noticed the difference. That flag is a small thing. But it points at something large. ## Three things, not one Most people experience logging in as a single ritual. You open a site, you prove who you are, you are in. But there are actually three distinct layers, and conflating them is how the weak spot stays hidden. The first is the account: your identity registered in a system. Creating one is often deliberately friction-heavy — email verification, phone numbers, sometimes document checks. That friction is intentional. The account is meant to mean something. The second is authentication: proving, each time, that you are the account holder. A password, a code sent to your phone, a biometric scan. This is the gate ritual, and it is where the industry puts most of its visible effort. The third is the session cookie: the lightweight token the service issues once you have passed the gate. From that point on, the cookie does the work. You stop re-proving yourself at every door. The cookie says: this person already checked in. ## The key card Think of checking into a hotel. You book a room, you show your passport at reception, and then you get a key card. From that point on, the card does the work. It opens your room, the gym, the parking barrier. Nobody asks for your passport again until checkout. The elaborate front-desk ritual gives you a sense of security. But what actually carries you through the building is a small piece of plastic with a magnetic stripe. Session cookies work the same way. You create an account, you authenticate, and the service issues a token. That token is your key card. And here is what I did not fully appreciate until recently: that card often sits as a plain file on your disk, accessible to any tool running on your machine. yt-dlp did not break into anything. It just picked up the key card I had left on the table. 🍪 Session cookies are stored locally by your browser. On most systems, tools with user-level access can read them without elevated permissions. Logging out of a service invalidates the server-side session, which is one reason it matters more than just closing the tab. ## The asymmetry Platforms invest heavily in the gate. Two-factor authentication, device recognition, login anomaly detection. The entry experience has become genuinely robust. But once the cookie is issued, many platforms let it run for weeks or months. Some bind it to a domain, few bind it to a specific device or IP address. That is the delegation problem stated differently: authentication systems were built around a single person at a single keyboard. The session token they produce was never designed to be transferred, borrowed, or held by an automated tool. Yet that is exactly what is happening, quietly, every time a CLI tool, a script, or an AI agent acts on your behalf using credentials you already established. ## What this changes For me, this sharpened two things. The first is personal. I feel less secure online than I did before understanding this. Not because anything has changed in the world, but because I now see where the actual surface area is. The gate is solid. The key card is lying on the table. The second is professional. Delegation across digital systems has always seemed technically hard. Giving an AI agent or a colleague selective access to act on your behalf without sharing your credentials felt like a problem requiring elaborate infrastructure. It still does, done properly. But the session cookie shows that a cruder form of delegation is already available, already working, and mostly invisible. That is not reassuring. It is clarifying. The elaborate check-in was never the whole story. It was the part we were meant to notice.

Stel je opent je analytics en ziet tienduizend pageviews. Goed gevoel. Maar wat als die tienduizend maar een fractie is van wat er werkelijk gebeurt, en de rest zich buiten je dashboard afspeelt? Dat is geen hypothese. Het is wat ik zie in mijn eigen logs. ## Onzichtbaar verkeer Menselijke bezoekers met een browser zijn op een gemiddelde dag rond de tien procent van mijn totale verkeer. De rest is een mix van indexering, geautomatiseerde retrieval en AI-gedreven toegang. Wanneer iemand een AI-assistent een vraag stelt, gaat dat systeem vaak in real time op zoek naar actuele informatie. Het bezoekt je site, leest je content, en verdwijnt weer. Geen sessie in Analytics. Geen bounce rate. Niets. Je bereik is dus veel groter dan je denkt. Maar ook veel minder transparant. ## De onverwachte held En dan komt RSS om de hoek. **Really Simple Syndication** , een formaat uit de begintijd van het web, dat de meeste mensen hadden afgeschreven toen Google Reader in 2013 sloot. Geen JavaScript. Geen opmaak. Gewoon inhoud, metadata en een tijdstempel, netjes verpakt in een gestructureerd bestand. Precies wat een AI-systeem zoekt. RSS was ontworpen voor mensen die controle wilden over wat ze lazen, zonder een website te hoeven bezoeken. Het blijkt dat geautomatiseerde systemen precies hetzelfde willen, om precies dezelfde redenen. De feedreader van 2005 en de AI-assistent van 2025 stellen identieke eisen aan een inhoudsstroom: gestructureerd, actueel, zonder ruis. Een goed onderhouden RSS-feed maakt je content vindbaar en bruikbaar voor deze nieuwe generatie lezers. Zonder nieuwe technologie. Het formaat was er al. ⚠️ RSS is XML-gebaseerd. Elke entry bevat een titel, link, publicatiedatum en meestal een samenvatting of volledige inhoud. Controleer of jouw website of blog een actieve RSS-feed heeft, de meeste platformen genereren die automatisch. ## Grip op het onzichtbare RSS staat niet alleen. Markdown, llms.txt, gestructureerde metadata: het zijn stille schakels in een infrastructuur die je bereik bepaalt zonder dat je het ziet. De meeste ondernemers denken aan SEO als ze aan vindbaarheid denken. Maar vindbaarheid verschuift. Zoekmachines worden gesprekspartners. Indexering wordt retrieval. Wie daar grip op wil houden, begint verrassend genoeg bij de oudste gereedschappen. Niet omdat ze nostalgisch zijn, maar omdat ze precies goed genoeg zijn voor wat er nu van ze gevraagd wordt. SPONSORED Hiveminds Lens analyseert hoe zichtbaar je bedrijf is in ChatGPT, Claude, Google Gemini en Perplexity. We testen relevante prompts, vergelijken je met concurrenten en laten zien welke technische, content- en authority-signalen ontbreken. Gratis snapshot

At 07:43 UTC this morning, something hit my site. Not a flood in the security sense: no alarms, no WAF triggers, nothing in the human analytics. Just a number that shouldn't have been that high, on a chart I check more out of habit than worry. Four hundred requests in sixty seconds. Then back to normal. My human-facing analytics saw nothing. That absence is itself a clue. ## Two layers Most publishers have one view of their traffic: the analytics dashboard. It shows pageviews, sessions, referrers, the countries their readers come from. It tells them what humans did. What it doesn't show is everything else. And everything else, it turns out, is interesting. I run a Cloudflare Worker that logs every request to a D1 database before passing it along. Every request: human, bot, crawler, scraper, agent. The Worker tries to classify each one, matching user-agent strings against a database of known bot signatures. What it can't classify, it logs as unknown. That second layer is where the detective work happens. ## Reading the evidence The 07:43 spike broke down like this: 202 of those 400 requests were for `.md` paths. Paths like `/when-bots-become-readers.md`, `/web-traffic-and-the-rise-of-llms.md`, `/measuring-traffic-machines-bots.md`. The Worker classified 192 of them as human, because the user-agent strings looked like browsers: Chrome 138, Firefox 115, Edge 114. Perfectly formed, perfectly plausible. But one user-agent hit 47 different `.md` paths in sixty seconds. Another hit 37. Both from the same two browser fingerprints, distributed across 25 countries. Then the robots.txt requests: 116 of them in the same minute. That's a preflight pattern, a swarm checking the rules before it reads the content. Chrome 114 hasn't been a current browser for a long time. Neither has Firefox 115. These are frozen strings, a signature of bot infrastructure that picks a browser version and pins it, never updating. The User Agent (UA) looks human. The behaviour doesn't. The conclusion assembled itself: a distributed scraper, running across a proxy network or botnet, using spoofed browser identities to avoid classification. Evasive, coordinated, and genuinely clever. ## The twist Here's where the detective story gets uncomfortable. Those `.md` endpoints don't exist by default in Ghost. I added them. A few months ago, as an experiment: serve each post as clean Markdown alongside the HTML version, reference them in `llms.txt`, see what happens. The idea was to make the content easier for AI systems to consume. Structured, clean, no JavaScript noise. The scrapers found them almost immediately. So did legitimate AI users, ChatGPT-User and OAI-SearchBot among them, reading the same paths through the same door. I set the bait. They smelled it. ## The idée fixe The reflex response to a traffic spike like this is defensive. Block the IPs, rate-limit the endpoint, add a CAPTCHA, harden the WAF. There is an entire industry built around that reflex. It rests on a premise worth examining: that keeping machines out is possible, and that it is worth the effort. Neither is quite true. A scraper that can distribute across 25 countries and rotate frozen browser UAs is not stopped by a robots.txt entry or a Cloudflare rule. It routes around friction the way water routes around a stone. And the content, once published on the public web, is going to be consumed by machine pipelines whether or not you make it easy. The more interesting question is what you can learn by watching it happen. The spike told me that my Markdown experiment is working, in the sense that it is attracting exactly the traffic it was designed for. It told me that the machine layer of the web is active, distributed, and more sophisticated than most people assume. It told me that the gap between what human analytics show and what the full request log shows is where the real picture lives. Blocking that traffic would have closed the window. Watching it left it open. ## Signal in the noise The thing about running a second logging layer is that most of what it catches is unremarkable. Googlebots, Bingbots, ChatGPT-User ticking through recent posts, the usual crawl of SEO tools and RSS readers. Noise. But the noise is the baseline. Without it, the 07:43 spike is invisible. With it, you can ask: what's different about this minute? Why these paths? Why this many countries? Why frozen UAs? The detective work is in the filtering, not the blocking. 🗒️ The .md endpoints at hoeijmakers.net are intentional. Each post is available as clean Markdown alongside the HTML version. The llms.txt file indexes them. This is an ongoing experiment in machine-readable publishing. * * * I Thought I Was Optimising for SpeedBuilding a caching layer for my blog turned into something else: a window into who actually reads it, and what reading even means now.Rob HoeijmakersRob HoeijmakersThirty Years of Caching, Sorted in an AfternoonHTTP caching never quite made sense, until AI tools made it legible enough to actually implement. And the reason it finally mattered: the audience had quietly changed.Rob HoeijmakersRob HoeijmakersMy Visitors Are Not All Human. That Is Fine.I built a traffic dashboard for my own site. What I found wasn’t alarming, it was interesting. A publisher’s notes on bots, borrowed identities, and editorial agency.Rob HoeijmakersRob HoeijmakersGuests That Should BehaveBots are modern guests. Most are welcome. But when they arrive in disguise with a real browser and spike your analytics, hospitality has limits. A Cloudflare story.Rob HoeijmakersRob HoeijmakersMarkdown, the WD-40 of Digital InformationMarkdown has barely changed in twenty years while everything around it was rebuilt. That’s not luck. It’s what happens when a format finds the exact sweet spot between content and meaning.Rob HoeijmakersRob Hoeijmakers

Mijn analytics lieten een goede week zien. Totdat ik de ruwe logs erbij pakte en begon te tellen. De helft van mijn bezoekers was geen mens. Dat klinkt alarmerend. Maar het is gewoon de werkelijkheid van elke website in 2026. En wie dat weet, staat er anders in dan wie het niet weet. ## Twee soorten verkeer Je website trekt bezoekers die lezen. En bezoekers die niet lezen: zoekmachines die indexeren, AI-systemen die content ophalen voor hun kennisbank, beveiligingsscanners die zoeken naar kwetsbaarheden, en geautomatiseerde tools die jij niet hebt uitgenodigd. Al dat verkeer komt samen in een getal in je dashboard. Plausible, Google Analytics, wat je ook gebruikt: het telt alles op. Zolang je niet verder kijkt, zie je een vertekend beeld van wie je publiek eigenlijk is. Het nuttige onderscheid is dit: er zijn geautomatiseerde bezoekers die jouw content ergens doorsturen naar echte mensen, en er zijn geautomatiseerde bezoekers die er niets zinnigs mee doen. Een AI-crawler van OpenAI of Anthropic valt in de eerste categorie. Die haalt je pagina op zodat een gebruiker van ChatGPT of Claude straks een antwoord krijgt dat deels op jouw schrijven is gebaseerd. Een vulnerability scanner die zoekt naar webshells op een server die niet de jouwe is, valt in de tweede. ****Let op:**** Een AI-crawler haalt jouw pagina op voor een model of index. Een AI-gebruiker is een persoon die via ChatGPT of Claude straks jouw content tegenkomt, zonder jouw site zelf te bezoeken. De crawler bezoekt jou. De gebruiker niet altijd. ## Waarom het ertoe doet Als je alleen naar menselijke bezoekers kijkt, mis je een groeiend deel van je bereik. Wie content schrijft, schrijft die inmiddels voor twee soorten lezers: mensen die klikken, en systemen die indexeren voor mensen die straks vragen stellen. Dat zijn andere eisen. Een AI-systeem geeft niets om je lay-out. Het wil schone tekst, duidelijke structuur, en snelle laadtijden. Precies wat menselijke lezers ook willen, maar om andere redenen. Goede caching en consistente headers zijn hierbij niet alleen techniek, het is onderdeel van je publiceerstrategie. Tegelijkertijd wil je de andere categorie, de automation die niets bijdraagt, wel uit je cijfers houden. Niet uit angst, maar omdat het ruis is. Wie zijn publiek niet kent, kan er ook niet voor schrijven. ## Wat je kunt doen Je hoeft geen eigen Cloudflare Worker te bouwen om dit beter te begrijpen. Maar een paar dingen helpen. Kijk naar je server logs, niet alleen naar je analytics dashboard. Ze vertellen verschillende verhalen. Begrijp wat de bekende crawlers zijn: Googlebot, Bingbot, de AI-crawlers van de grote modellen, de SEO-tools. Die zijn welkom. Ze helpen je bereik vergroten via andere kanalen dan directe klikken. Blokkeer bewust wat je wilt blokkeren, niet als reflex. Een robots.txt die AI-crawlers weert, is een keuze die je kunt maken, maar het is een keuze met gevolgen voor je zichtbaarheid in AI-gestuurde antwoorden. Weten wat er rondloopt geeft je die keuze terug. Het merkwaardige van dit alles: ik begon met een technisch project, snelheid optimaliseren, en eindigde met een scherper beeld van mijn publiek dan ik in jaren had gehad. Niet ondanks de bots. Doordat ik ze kon labelen. Voor de technische achtergrond: * Thirty Years of Caching, Sorted in an Afternoon * My Visitors Are Not All Human. That Is Fine. * I Thought I Was Optimising for Speed

I have a jar of screws on my workbench. For years, I would fish through it looking for the right size, usually not finding it. Last week I sorted them, by type, by thread, by length. I used ChatGPT to help: photographed a handful, asked what I was looking at, got the taxonomy straight. Once I could name them, I could organise them. You can only sort what you understand. HTTP caching was my jar of screws. ## Thirty years of fog I have been building for the web since the early nineties. Caching was always there, somewhere in the background, doing something. I knew enough to be aware of it, not enough to actually control it. Cache-Control headers, TTL values, edge behaviour, the difference between what a CDN caches and what a browser holds, what gets invalidated when and why. Every time I approached it seriously, I ran into a wall of context I did not quite have. The documentation exists. The concepts are not secret. But caching is one of those domains where the gap between understanding the vocabulary and being able to apply it correctly is surprisingly wide. I would read, nod, implement something plausible, and move on with lingering doubt. This year, working with Claude, that changed. The parallel is closer than it sounds. I had the pieces in front of me for years. What I was missing was someone to explain what I was looking at. ## New instruments We went through the whole thing together. What my Cloudflare Workers were actually doing. What headers were being sent and why. What a browser would cache versus what the edge would cache. Where the inconsistencies were. What a coherent strategy would look like for a site like mine: a moderate personal blog with a global readership, running on Ghost, served through Cloudflare. It took an afternoon. Not because the subject got simpler, but because I had, for the first time, an instrument that could hold the full complexity with me. Ask a question, get an answer calibrated to my exact setup, follow a thread, revise, implement, check. The back-and-forth that used to require either a specialist or weeks of trial and error compressed into something manageable. The result was a caching strategy I actually understand. Headers that mean what I intend. Edge behaviour that is consistent. Rules I can read back and explain. ## The audience had already changed The reason it finally felt urgent was not vanity metrics or pagespeed scores. It was a shift in who was actually reading. Human visitors are still there. But a growing share of traffic to a site like mine now comes from crawlers: search indexers, AI training pipelines, retrieval systems that serve content to agents rather than browsers. These systems do not render pages. They do not wait for JavaScript. They send a request, receive a response, and move on. For them, caching is not a convenience. It is the primary mechanism that determines cost, latency, and reliability of access. If you care about how your content moves through the world now, including through AI systems, you have to care about caching. Not as a performance optimisation for human browsers, but as infrastructure for machine readership. That reframing changed what I was optimising for. HTML cached at the edge, globally, with consistent headers and predictable expiry. Not because I expect a person in Singapore to shave 200ms off their pageload, but because the next request for that page is more likely to come from a retrieval system than a browser, and the request after that, and the one after that. Cloudflare tiered caching on top of Ghost blog. ## What the tools made possible The caching itself is not new. The concepts are decades old. What changed is that I could finally see the system clearly enough to shape it. With the right instrument, a domain that had been opaque for thirty years became workable in a single session. That is not a small thing. There are other jars of screws on the workbench. 🗒️ The caching strategy described here was implemented using Cloudflare Workers, Cache Rules, and D1 for request logging. A public cache-stats dashboard shows the current breakdown of traffic by type: human, AI crawler, SEO crawler, and unknown. * * * My Visitors Are Not All Human. That Is Fine.I built a traffic dashboard for my own site. What I found wasn’t alarming, it was interesting. A publisher’s notes on bots, borrowed identities, and editorial agency.Rob HoeijmakersRob HoeijmakersWhen Bots Become Readers: Publishing in the Age of AI CrawlersListening to Matthew Prince on Azim Azhar’s podcast made me reflect on who actually reads my blog. People (like you), machines, or both.Rob HoeijmakersRob HoeijmakersThe End of Google Search (as we know it)Google didn’t warn me. It just erased my blog. What looked like a bug turned out to be a glimpse into the future of search—and it’s not built for us anymore.Rob HoeijmakersRob Hoeijmakers

I had, without quite deciding to, written ChatGPT off for images. The earlier output felt like a tool optimised for fantasy illustrations, product mockups, and memes. Useful for some things. Not for the way I work. My interest in visualisation has always been functional: diagrams, schematics, process maps. Images that explain something rather than represent something. For that category, AI image generation had consistently disappointed. Then last week I needed to understand how a washing machine actually works internally. The inner drum, the outer tub, the bearings, the door seal, where the shaft exits the housing. Text descriptions get you some of the way. A diagram would get you further. So I asked ChatGPT Images 2.0, running on the new GPT Image 2 model, to generate one. What came back made me revise my assessment. ## What arrived The output was not a pretty picture of a washing machine. It was a coherent technical infographic: multiple views of the same machine, labelled components, readable typography, front and rear logic, the drive belt and motor in correct relation to each other. It carried layout hierarchy. It had editorial polish. It looked less like image generation and more like a service manual that had not yet been published. The surprise was specific. A washing machine is a good test precisely because it has no aesthetics to fall back on. The image is either mechanically correct and useful, or it is not. There is no artistic merit to hide behind. And this one passed. ## Why this category matters The images I care about are not decorative. Every organisation explains something: how a process works, how a system fails, what happens in a given sequence. For years, turning that complexity into a clear visual required time, a specialist, and budget. The bottleneck was not the knowledge, it was the production. GPT Image 2 is built around the idea that the model thinks before it draws, reasoning about what an explanation actually requires before generating anything. Text inside images, historically the biggest weakness of these models, has improved substantially: labels, signage, UI elements, correct spelling and consistent spacing. The result is a model that does not just render, it plans. That is the shift. Not better art. Better explanation. 💡 GPT Image 2 launched on April 21, 2026. It introduced thinking mode, web search during generation, and the ability to produce up to eight consistent images from a single prompt. Available to all ChatGPT users; extended outputs require a paid plan. ## The category I had dismissed Many people still assess AI image tools by the use cases that dominated the first wave: avatars, memes, synthetic photography, visual gimmicks. That assessment is not wrong, it is just stale. The more consequential category is knowledge visuals: images that help a professional understand something faster than prose alone. Technical diagrams, process flows, onboarding schematics, infrastructure maps. The kind of thing that organisations need constantly and produce slowly. If that gap is now measurably shorter, the operational implication is significant. Not because designers are replaced, but because explanation becomes cheap enough to happen more often. I only wanted to know how my washing machine works. The diagram that came back raised a different question: what else is now explainable in minutes that previously required a budget line?

Wilhelm von Humboldt is one of those figures I knew only by association: the brother, the geographical namesake, the name on the university. The actual ideas, I had missed. A piece from the Cosmos Institute put Humboldt's concept of _Bildung_ in front of me. The argument is simple and uncomfortable: education shaped by Humboldt was about forming a whole person, not training a function. The modern professional world, by contrast, has spent a century doing the opposite. We have been very good at turning people into functions. ## The Bildung idea _Bildung_ is hard to translate, which is part of why it stayed in German. It means something like self-cultivation through engagement with the world: art, science, language, culture, other people. German has a second word for the other model: _Ausbildung_ , vocational training, formation toward a function. The prefix _aus_ signals completion, extraction, being shaped into a finished product. One word for becoming a person. One word for becoming a role. The goal of _Bildung_ is not competence in a domain. The goal is a richer, more capacious person. Humboldt believed the university existed to develop that person, not to fill a role. This is not naive idealism. Humboldt was a state official. He designed the University of Berlin in 1810 with this philosophy embedded in the structure: research and teaching together, students as active participants, knowledge pursued for its own sake. The model spread across Europe and shaped what we call research universities. ## The function trap What the Cosmos Institute piece points at, and what I find genuinely interesting, is that AI is now exposing a tension that was always there. When knowledge work was scarce and complex, being a good function was valuable. You knew the law, the process, the domain. That knowledge was your identity, your leverage, your job security. AI is eroding the scarcity. Not all of it, and not evenly, but the trend is clear enough that the question becomes: what remains when the function is cheaper to automate than to hire? The Humboldt answer, if taken seriously, is that what remains is the person. The capacity to make judgements, to hold complexity, to bring a formed perspective to an ambiguous situation. None of those come from function-training. They come from exactly the kind of broad, messy, integrative engagement that _Bildung_ describes and modern professional education systematically devalues. ## What I do not know I want to be honest about the limits of what I am saying here. I have not read Humboldt, I have encountered the idea. I am not a philosopher of education or a historian of the university. What I can say is that the frame feels generative: it names something I have been sensing without having the word for it. The risk of reaching for Humboldt now is that it becomes a nostalgic move, a way of dressing up the resistance to change in the language of classical education. The universities Humboldt influenced are not obviously doing _Bildung_ today. They are credentialing machines with research departments attached. The more interesting question is whether the AI transition creates conditions where something like _Bildung_ becomes practically useful again, not as philosophy, but as a competitive model for how people and organisations develop. Whether that is inspiration or a warning, I genuinely do not know. 📎 The Cosmos Institute piece that triggered this: "You Are Not a Function" * * * Vision, Judgement, Creativity: Reclaiming Agency in the Age of AIFrom analysis paralysis to agency: why vision, judgement and creativity matter most in an AI-shaped world.Rob HoeijmakersRob Hoeijmakers

Staatssecretaris Willemijn Aerdts (Digitale Economie en Soevereiniteit) communiceert voortaan met een handvol ambtenaren via een Europese berichtenapp. Welke? Dat wil haar woordvoerder niet zeggen. Het initiatief staat "nog in zijn kinderschoenen." Toch is de mededeling zelf veelzeggend: de Nederlandse overheid vindt haar afhankelijkheid van WhatsApp en Signal een risico. Dat is laat, eerlijk gezegd. Frankrijk gebruikt Tchap al sinds 2019, Duitsland heeft de Bundeswehr al jaren op een eigen berichtenapp, België lanceerde vorige maand Beam voor 750.000 ambtenaren en militairen. Maar beter laat dan nooit, en de richting is helder. ### Groter dan een overheidsproject De reden dat overheden dit nu serieus nemen heeft minder met privacyideologie te maken dan met geopolitiek. Trump is terug, tech wordt ingezet als machtsmiddel, en de vraag "kunnen we morgen nog bij onze data?" is geen paranoia meer. Een enquête van SWG uit januari liet zien dat 59% van de Europeanen het blokkeren van digitale diensten door de VS al als een reëel risico ziet. 86% vindt het op zijn minst plausibel. Voor overheden is dat een legitieme afweging. Maar het raakt ook bedrijven, alleen merken die dat minder snel. ### De strijd om de AI-laag Terwijl overheden zoeken naar alternatieven, speelt er tegelijk een juridisch gevecht dat dichter bij de ondernemer zit. De Europese Commissie heeft Meta afgelopen weken voor de tweede keer een "charge sheet" gestuurd, ditmaal over de tarieven die Meta wil rekenen aan externe AI-aanbieders zoals OpenAI en Anthropic op het WhatsApp Business-platform. De aanleiding: Meta verbood die externe aanbieders aanvankelijk gewoon. Toen de EU dat aanvocht, introduceerde Meta een nieuw model waarbij externe AI-chatbots 5 tot 13 cent per bericht betalen. Meta AI zelf, de eigen assistent, valt buiten dat tarief. De EU vindt dat dit neerkomt op hetzelfde: een verkapte ban die Meta's eigen AI bevoordeelt. Dat klinkt als een aangelegen zaak voor techjuristen, en voor de meeste ondernemers is het dat ook direct. Wie via de WhatsApp Business API een eigen chatbot draait, via een BSP zoals Enreach of Twilio, betaalt al gesprekskosten aan Meta via zijn provider. Dat verandert niet door dit conflict. Het tariefgeschil gaat over iets anders: grote AI-aanbieders als OpenAI en Anthropic die WhatsApp als distributiekanaal willen gebruiken voor hun eigen producten. Meta wil die toegang belasten; de EU vindt dat een verkapte ban. De indirecte impact voor bedrijven is subtieler. Als Meta de spelregels voor platformtoegang verder aanscherpt, of als BSP's onder druk komen te staan, merk je dat uiteindelijk ook als eindgebruiker. Niet morgen, maar de richting is gezet. Meta AI in WhatsApp ### Wat je er nu al uit kunt lezen WhatsApp is geen neutraal kanaal meer. Het is een platform met eigen commerciële belangen, Europese juridische druk, geopolitieke fricties en een AI-strategie die de concurrentie actief benadeelt. Dat is niet een reden om er morgen mee te stoppen, want de penetratie in Nederland is boven de 85% en je klanten zijn er gewoon. Maar het is wel een reden om niet alles op die ene infrastructuur te bouwen. Bedrijven die hun klantcommunicatie volledig op WhatsApp hebben ingericht, hebben stilzwijgend een afhankelijkheid gecreëerd. Een beleidswijziging, een account-blokkade, een tariefaanpassing: ze komen zonder aankondiging. Wat de overheid nu formuleert als een soevereiniteitsvraag, is voor bedrijven eigenlijk gewoon risicobeheer. De Nederlandse pilot is klein en voorzichtig. Maar de beweging die erachter zit, is dat niet.

A friend asked which Claude subscription he should get. The pricing page left him more confused than when he started. I can't tell him what's right for his situation, but I can tell him what I do. ## The setup I run a Pro subscription. That covers the bulk of what I need: writing, analysis, research, structured thinking, light agentic work. I hit the daily usage limit occasionally. The weekly limit, rarely. When I do hit the daily limit, Claude switches to prepaid credit automatically. No interruption, no decision to make. I top up the balance now and then, and the buffer keeps working in the background. Total monthly cost stays well below the Max plans, because I only pay overage when I actually need it. That asymmetry matters. Max makes sense if you're hitting limits constantly or need guaranteed capacity. For the way I work, occasional overage is cheaper than a high fixed floor. ## Model choice Sonnet or Opus makes a real difference here too. Opus is the more capable model and costs significantly more per conversation. I use Sonnet for most things, and it handles the bulk of what I do well enough. Opus comes out when the problem genuinely requires it. The instinct to always reach for the most powerful model is understandable. It's also where costs quietly compound. ## The invisible meter The trickier part is that token usage isn't visible while you're working. It's a bit like bandwidth: you don't think about it until something trips. A conversation feels like a conversation, not like a metered resource ticking down. It gets more pronounced when you run things in parallel. Multiple agentic sessions, several long document conversations open at once: usage can spike in ways that aren't obvious until you've hit a limit or a bill arrives. The more autonomous the setup, the less usage feels connected to the work you're actually doing. There are probably tools and commands available to check usage more precise but for me the Claude usage page under settings is the dashboard. ## The API side I also keep a small prepaid credit balance for the API, which is separate from the claude.ai subscription. I use it for tools I've built, and occasionally for Claude Code in the terminal. The two balances don't talk to each other: a Pro subscription on claude.ai doesn't cover API usage, and API credit doesn't unlock Pro features. 💡 Claude is one of several AI tools I use daily. ChatGPT plays a different role in my workflow: more thinking and cognitive exploration, where Claude is more about doing and building. I wrote about that distinction here → "ChatGPT thinks with you, Claude builds with you" ## Plans as a moving target My friend is on Max, partly because he's waiting for the Word plugin, and partly because he suspects that being in a plan early pays off as Anthropic adds features over time. That last point is probably right. Anthropic does adjust what's included in plans as products mature, and being an established subscriber tends to work in your favour. For him, Max makes sense. For me, Pro plus a small credit buffer is still the more cost-efficient fit. The right answer depends on how you work, not on which plan looks most complete on a pricing page. Start with Pro. See where the limits actually land for your patterns. Add a credit buffer so a daily ceiling doesn't stop you mid-afternoon. Reassess after a month. * * * ChatGPT Thinks With You. Claude Builds With You.I switched from ChatGPT to Claude. The difference isn’t the model. It’s the shift from cognitive support to digital crafting, and what that makes possible.Rob HoeijmakersRob Hoeijmakers

Een vriend vroeg me welk Claude-abonnement hij moest nemen. Hij had de prijspagina van Anthropic bekeken en stond verwarder dan daarvoor. Ik snap waarom. De indeling van Claude-abonnementen is niet ontworpen rondom een herkenbaar werkpatroon. Pro, Max, Teams — het zijn lagen, geen verhalen. Toch is de keuze minder ingewikkeld dan hij lijkt, als je weet waar je op moet letten. ## Pro als basis Pro is het instapabonnement voor dagelijks gebruik. Schrijven, analyseren, onderzoek, lichte automatisering. De meeste ondernemers die Claude serieus inzetten, komen hier mee door. Er zit een dagelijks gebruikslimiet op, en een wekelijks limiet. Wie niet constant tegen die grenzen aanloopt, heeft aan Pro genoeg. Wat veel mensen niet weten: bij Pro kun je een apart tegoed opladen als aanvulling. Als je de dagelijkse limiet bereikt, schakelt Claude automatisch over op dat tegoedbudget, zonder onderbreking. Je betaalt dan per verbruik, bovenop het vaste maandbedrag. Voor wie af en toe een drukke dag heeft maar niet structureel, is dat goedkoper dan doorupgraden naar Max. ## Max voor constant gebruik Max-abonnementen zijn er in twee varianten, die van elkaar verschillen in hoeveel berichten je per dag kunt sturen. Wie dagelijks intensief werkt met Claude, meerdere projecten tegelijk bijhoudt of lange documentsessies heeft, merkt bij Pro de grenzen sneller. Max geeft meer capaciteit voor een vaste prijs. Voor wie wacht op specifieke features die nog komen, of denkt dat een vroeg abonnement op termijn voordeel oplevert, is Max ook een overweging. Anthropic heeft de gewoonte om Pro-features eerst beschikbaar te stellen aan betalende gebruikers, en hogere plannen doorgaans eerder. ## Modelkeuze maakt ook uit Binnen elk abonnement kies je welk model je gebruikt. Sonnet en Opus zijn de twee hoofdopties. Opus is het krachtigere model en kost meer per sessie. De reflex om altijd voor het krachtigste te gaan is begrijpelijk, maar voor de meeste taken is Sonnet voldoende. Het verschil merk je pas bij complexe, meerlagige vraagstukken. Het is niet de instelling die het meeste kost, maar de gewoonte om hem te gebruiken voor dingen die niet zo zwaar hoeven te zijn. ## API-credits zijn apart Wie Claude ook via de API wil gebruiken, voor tools, scripts of integraties, betaalt daarvoor apart. API-credits staan los van het claude.ai-abonnement: het ene dekt het andere niet. Dat is verwarrend totdat je beseft dat het twee aparte producten zijn: de webinterface enerzijds, het platform voor ontwikkelaars anderzijds. Voor ondernemers die niets bouwen en Claude uitsluitend via de chat gebruiken, is dit niet relevant. ## Waar het op neerkomt Gebruik Claude een paar weken op Pro. Let op waar de grenzen vallen, niet op papier, maar in jouw werkdag. Als je regelmatig halverwege de middag stokt, schakel dan op. Als je dat nooit hebt, blijf dan zitten. De beste indeling staat niet op de prijspagina. Die staat in je eigen gebruik. 💡 Claude heeft ook een Teams-abonnement voor organisaties. Dat voegt beheer en gedeelde projecten toe. Voor zelfstandigen en kleine bedrijven is het persoonlijke Pro-abonnement met losse credits in de meeste gevallen de kostenefficiëntere keuze, maar ik heb er persoonlijk geen ervaring mee. * * * Van gratis naar betaald: hoe kies je het juiste AI-model voor je organisatieVeel bedrijven experimenteren met gratis AI-modellen, maar voor structureel gebruik gelden andere eisen. Wat zijn de juiste keuzes voor Nederlandse organisaties?Chat voor bedrijvenRob HoeijmakersDiepgaand onderzoek versus agentmodus in ChatGPTBeide modi binnen ChatGPT lijken te “denken”, maar één begint met “doen”. Een verkenning van de stille overgang van analyse naar operatie in alledaagse AI-tools.Chat voor bedrijvenRob HoeijmakersWat ChatGPT Personal en Business echt onderscheidt: context, geheugen en ontwerpChatGPT Business werkt fundamenteel anders dan de persoonlijke variant. Niet qua functies, maar in hoe context en geheugen bewust zijn afgebakend.Chat voor bedrijvenRob Hoeijmakers

The announcement slide said "NO TRACEABILITY" in large capitals on a blue background ringed with gold stars. The Threads comment section replied: "How do we verify children, ah yes let's make big database of all people." The architecture is the opposite of a database. The communication guaranteed nobody would notice. Left LinkedIn, right Threads. ## What they built The Age Verification App is a "mini wallet," built on the same technical foundation as the forthcoming EU Digital Identity Wallet. When a platform requests age verification, the app generates a cryptographic proof derived from a credential issued by a national identity authority. The platform receives one binary answer: over 18, or not. No name, no date of birth, no address. The underlying identity data never leaves your device. The same mechanism works offline. At a festival gate, a bar, a tobacconist. You present your wallet, the venue's device gets a yes or no, and sees nothing else. Most adults have handed a passport to a bouncer to prove they are old enough to enter, giving a stranger full sight of their identity because the system had no finer instrument. This changes that. The credential is cryptographically bound to the issuing device. It cannot be copied or handed to someone else; the proof would fail. The weak point is onboarding: if someone sets up a wallet using another person's ID document, the credential reflects that person's age. A process problem, not a cryptographic one. 💡 The blueprint is open source at ageverification.dev. Six member states including France, Spain, and Denmark are currently in pilot. A second version, released October 2025, adds passport and ID card onboarding alongside eID. ## Two failures, one root cause The Commission made two distinct mistakes, and they compound each other. The first is aesthetic. Blue backgrounds, gold stars, caps-lock claims. This is the design language of an institutional poster, activating exactly the distrust it was meant to address. Design communicates before words do. A system built on cryptographic privacy, dressed in the visual language of state surveillance, loses the argument before it starts. Estonia's e-Residency programme, adjacent identity infrastructure, answered three questions before anything else: what is it, who is it for, how do I get one. The second is structural. "The European Age Verification App is ready" is product language. It implies: find it, download it, use it. What actually launched is an open-source blueprint that member states can adapt, currently in pilot in six countries, dependent on national identity infrastructure that varies by country, with no App Store link, no rollout date, no answer to the question any citizen would ask: what do I actually do with this, and when? Both failures share a root. The Commission framed this entirely as a child protection measure. "It is for parents to raise their children. Not platforms." Every word is defensive. It excludes every adult who benefits from not uploading a passport scan to access a service, or not handing their full identity to a bouncer at a festival. The genuine upside, comfort, convenience, less data exposed in daily life, went unspoken. It is for parents to raise their children. Not platforms. The European Age Verification App is ready. It will allow users to prove their age when accessing online platforms. Just like shops ask for… | European Commission | 48 commentsIt is for parents to raise their children. Not platforms. The European Age Verification App is ready. It will allow users to prove their age when accessing online platforms. Just like shops ask for proof of age for people buying alcoholic beverages. And it ticks all the boxes: ✅ Highest privacy standards in the world ✅ Works on any device ✅ Easy to use ✅ Fully open source More info: https://lnkd.in/dXQeA6Kx | 48 comments on LinkedInLinkedInThierry Thevenet ## The real circumvention problem The Threads commenter who noted "VPNs are also ready" was pointing at something real, though not quite in the way they meant. The cryptographic proof cannot be spoofed or transferred. But a VPN that makes a user appear to browse from outside EU jurisdiction sidesteps the obligation to verify at all. The circumvention is not of the architecture; it is of the enforcement. This is a DSA problem, not a technical one. Whether platforms are compelled to integrate the system, and whether that compulsion reaches beyond EU borders, are open questions the announcement did not address. The blueprint closes one circumvention route. Jurisdiction closes another only if enforcement follows. ## The unlaunched product The technical work is done, and it is genuinely good. What has not started is the communication work: framing this around the adult user who wants to prove their age without surrendering their identity. That person also happens to satisfy the DSA, protect children, and reduce platform liability. The upside was always there. The Commission chose not to use it. A system designed to preserve privacy should not have to overcome the impression that it destroys it. That gap is not a technical problem. It is a choice. * * * Understanding Digital Identity in the EU: eID, eIDAS, and the EUDI WalletThis article serves as a reference point for further discussions on digital identity, authentication, and legitimisation in the EU.Rob HoeijmakersRob HoeijmakersWero: Europe’s new payment brandWero is Europe’s bid for payment sovereignty: pay across borders with just an email or phone number, without relying on US networks.Rob HoeijmakersRob Hoeijmakers

I read a news article last week about Amsterdam asking data centres to reserve space for European clients. Something bothered me about it. I opened a new document and started writing. Three days later I had notes for three separate pieces, none of them finished, and a growing suspicion that the thing bothering me was not what I thought it was. ## The split The first article was about category errors. The piece conflated continuity, confidentiality and security, three distinct problems that require different interventions. I wrote that one fairly quickly. It felt satisfying in the way that corrections always feel satisfying, precise, a little superior. I set it aside. The second article was about Amsterdam specifically. The city at the centre of European internet infrastructure, home to one of the world's largest internet exchange points, that spent years fighting data centres and now finds itself asking the ones it allowed to hold space for European clients. The irony was clean. Too clean, maybe. I set that one aside too. The third article was the one I actually wanted to write. It was about the NIMBY problem. How the cities most anxious about digital sovereignty are the ones most aggressively preventing the infrastructure sovereignty requires. How we export our data centres the same way we export our waste: to somewhere poorer, somewhere with weaker rules, somewhere that lacks the political capital to refuse. And how the people who campaigned against the incinerator keep producing rubbish, just without having to look at it. That one felt true. But it also felt incomplete. Because the NIMBY argument is still a solutions argument. It still assumes we know where we are going and just need to remove the obstacles. And I am no longer sure we know where we are going. ## The habit underneath Here is what I kept circling back to. Digital dependency on American infrastructure is not a procurement failure. It formed the way habits form: for real reasons, over time, because it kept making sense. The services were good. The prices were low. The integration was seamless. You did not end up here through weakness or inattention. You ended up here because it worked, and kept working, until the geopolitics shifted and suddenly the habit was visible in a way it had not been before. The recovery metaphor is useful up to a point. You make a policy. You change the default at each renewal. You take some things in-house. You support European alternatives even when they are not quite as good. I have done some of this in my own practice. Not the hard parts yet. The core productivity stack is still there. I tell myself I am waiting for the right moment, which is probably true and probably also a rationalisation. But recovery assumes you know what you are recovering toward. What does a healthy relationship with digital infrastructure look like? What does a European digital society actually feel like, cost, require? Those questions do not have clean answers. And the conversation keeps avoiding them, because solutions are easier to talk about than destinations. ## Where we were going We built the digital world we built because we were optimising for something. Speed. Convenience. Scale. Price. The platforms that won were the ones that removed friction fastest and charged least. We did not ask, collectively, what we were giving up or where we were heading. We just went. Individuals, companies, governments, all of us, at different speeds, in the same direction. Now the direction is contested. And the response, ten-year strategies, cloud alliances, rack space negotiations, is still being framed as a logistics problem. How do we get from here to sovereignty? What are the steps, the phases, the milestones? I keep wanting to ask a prior question. Sovereignty for what? A European digital society organised around what values, what trade-offs, what vision of what we owe each other and what we want to keep private and what we are willing to share? That conversation is not happening at the policy level. It is barely happening at the cultural level. And without it, the logistics question has no real answer, because you cannot plot a route to a destination you have not named. ## The unfinished document So I have three unfinished articles and this one, which is about why I could not finish them. I think the honest position is this. The sovereignty conversation is real and urgent and I take it seriously. I am also not sure the vocabulary we have for it is adequate to the actual problem, which is less about infrastructure than about intention. Less about where the servers are than about what kind of digital life we are trying to build, and for whom, and under whose rules, and whether we are willing to pay what it costs. I do not have an answer to that. I am not sure anyone does right now. But I think it is the question sitting underneath all the other questions, and I wanted to at least put it on the page. The three articles will probably get written eventually. This one needed to come first. * * * The Paradox of Sovereignty: Europe’s Search for Freedom in a Connected WorldThe EU talks about digital sovereignty as independence, yet its strength still depends on shared systems, U.S. clouds, and global code.Rob HoeijmakersRob HoeijmakersThe AI Continuity ProblemAI is load-bearing in my company. That creates a new kind of business risk: what happens when access breaks, gets priced out, or gets cut off?Rob HoeijmakersRob Hoeijmakers

I visited Leuven last year. Beautiful city, perfectly preserved, Gothic town hall, students everywhere. On its edge I walked past a modern building with the name imec on it. It looked serious. I had no idea what it was. Turns out it might be one of the most strategically important buildings in the global semiconductor industry. ## Nobody's chip company Imec employs researchers from over a hundred countries and holds no manufacturing lines of its own. It is, by design, nobody's chip company. That turns out to be the point. In 1984, the Flemish regional government did something unusual. It didn't attract a semiconductor manufacturer, or subsidise an existing one. It funded a research commons: a neutral laboratory where competitors could work on shared problems without handing advantage to a rival. The founding team were young researchers, several of them returned from Silicon Valley, backed by a government willing to think in decades rather than quarters. They called the ambition a SuperLab. It became imec. ## The paradox Most technology clusters form around an anchor company: ASML in Eindhoven, ARM in Cambridge. Leuven built the infrastructure first and let the ecosystem follow. Because Belgium had no dominant chip producer to protect, imec could be structurally neutral. Intel, Samsung, TSMC and others could collaborate there without the research benefiting a direct competitor. The absence of a national champion was the founding condition for a global one. ## Nine days ago That neutrality is now carrying serious weight. On March 18, imec announced the installation of the ASML EXE:5200: the most advanced lithography machine in existence, one of fewer than a dozen available anywhere in the world, valued at $400 million. It sits in imec's cleanroom in Leuven, available to the global ecosystem of chip makers, materials suppliers and research partners. Not owned by one company. Shared, on neutral ground, to accelerate what comes next. The machine targets sub-2nm chip development, territory that doesn't yet exist in commercial production. Imec receives the world’s most advanced High NA EUV system | imecA major milestone in propelling industry into the angstrom era.imec Forty years after the SuperLab idea, the logic holds. Europe doesn't manufacture the world's leading chips. But it owns two of the choke points that make those chips possible: ASML builds the only machines capable of printing them, and imec is where the industry learns how to use what comes next. ## Invention, not policy What did the Flemish government actually build in 1984? A research institute, formally. But the design choices, non-profit structure, university roots, enforced neutrality, long time horizons, were not incidental. They were the product. Imec works because of what it was prevented from becoming as much as what it was built to be. That's closer to invention than policy. The EU Chips Act is now directing billions toward semiconductor capacity across Europe, and imec sits at the centre of it. But imec's CEO Luc Van den Hove is more cautious than the policy framing suggests: allied collaboration, not autarky. The chip supply chain is too interdependent to be repatriated. What Leuven has can't be replicated by a funding announcement. It took forty years to build, and required above all the discipline not to turn it into something else along the way. 💡 Imec stands for Interuniversity Microelectronics Centre — in Dutch: Interuniversitair Micro-Electronica Centrum. Founded in 1984 as an autonomous non-profit, spun out of KU Leuven, one of Europe's oldest universities. They officially style the name in lowercase: imec. * * * How ZEISS and ASML Enable the Modern Chip IndustryFrom Dutch roots to German optics, ASML and ZEISS built a unique European stronghold in chipmaking. Where light carves the future.Rob HoeijmakersRob HoeijmakersStrengthening Europe’s Semiconductor Backbone: The Role of ESMCTSMC’s first European plant in Dresden, a €10 billion venture with Bosch, Infineon, and NXP, will enhance Europe’s semiconductor industry. Supported by the European Chips Act.Rob HoeijmakersRob Hoeijmakers

Laat ik beginnen met wat het níét is. EU Inc. is geen optie die je volgende week kunt gebruiken. Het is een voorstel dat nu formeel op tafel ligt bij de Europese Commissie. Dat is een stap verder dan een idee, maar nog ver verwijderd van een werkelijkheid. Dus als je nu nadenkt over internationaal groeien, een tweede entiteit, of een structuur die je makkelijker over grenzen kunt bewegen: dit is niet het antwoord voor vandaag. Maar het is wel een signaal dat de moeite waard is om te begrijpen. ## Waar het op inspeelt Als je ooit iets geregeld hebt in een ander EU-land — een klant in Duitsland bedienen via een lokale entiteit, een partner in België, een medewerker in Spanje — dan weet je hoe versnipperd dat is. Elk land heeft eigen registratieprocedures, eigen notarissen, eigen regels. EU Inc. wil daar iets naast zetten. Niet in plaats van, maar naast. Een 28ste optie, bovenop de 27 bestaande nationale vormen. Het idee: je kunt een Europese bedrijfsvorm kiezen die in alle lidstaten herkend wordt, zonder per land helemaal opnieuw te beginnen. ## Wat het wél en níét regelt Dit is het deel dat telt voor ondernemers. EU Inc. richt zich op registratie en rechtsvorm. Dat is de eerste stap. Maar de stappen daarna zijn een ander verhaal. Belasting, arbeidsrecht, je verplichtingen bij betalingsproblemen: die vallen buiten dit voorstel. Die blijven nationaal geregeld. Wat dat in de praktijk betekent: de administratieve drempel bij de start kan lager worden. De complexiteit daarna niet per se. Of dat genoeg is, hangt af van waar de wrijving bij jou zit. Als het starten van een entiteit in een ander land de bottleneck is, dan biedt EU Inc. iets. Als het gaat om belasting of arbeidsrecht, dan niet. ## Waarom het toch serieus te nemen is Er zijn eerder pogingen gedaan om iets vergelijkbaars te maken. De Societas Europaea bestaat al jaren, maar je komt hem in de praktijk nauwelijks tegen. Te complex, te weinig voordeel. Wat hier anders is: het is eenvoudiger geframed, en Ursula von der Leyen zet er haar naam onder. Dat geeft het politiek gewicht. Als het misgaat, is dat zichtbaar. Als het slaagt, ook. En de toon is anders dan bij eerdere voorstellen. Niet langer proberen om alles tussen lidstaten te harmoniseren, want dat is te traag en te ingewikkeld. In plaats daarvan: accepteer de versnippering, en bouw er een laag overheen die makkelijker te gebruiken is. Dat is pragmatisch. En voor bedrijven is pragmatisch meestal beter nieuws dan idealistisch. ## Wat je er nu mee kunt Plan er nog niets op. Maar als je groei over landsgrenzen heen overweegt, is dit het moment om te begrijpen wat de obstakels zijn. Niet voor EU Inc. specifiek, maar in het algemeen: waar zit de wrijving? Registratie, fiscaal, personeel? Als registratie jouw bottleneck is, houd EU Inc. dan in de gaten. Volg het proces. Het duurt jaren voordat dit wet wordt, maar de richting is nu gezet. En die richting zegt iets: Europa raakt ongeduldig met zijn eigen complexiteit. Dat is, voor ondernemers die internationaal willen, niet het slechtste nieuws. * * * Eén Europese onderneming (EU-INC)De EU werkt aan EU-INC: een optionele Europese bedrijfsvorm om grensoverschrijdend groeien eenvoudiger te maken. Nog geen wet, wel een duidelijke richting.Chat voor bedrijvenRob Hoeijmakers

Every now and then new hardware creates a small window for experimentation. Not a big project, just a moment to try features that were always there but never quite made it into daily use. I had two of those on my list: **Continuity Camera** , **Center Stage** , and something called **Desk View**. All three sit somewhere around the same idea: using camera technology a little more intelligently. The premise is straightforward. The camera in an iPhone is simply better than the cameras that used to be built into many Macs. Older iMacs and MacBooks often shipped with 720p webcams, mounted in a fixed position at the top of the screen. The result is familiar: a somewhat soft image, and a viewing angle that tends to look slightly up at you from below the chin. Using the iPhone changes both things at once. The image quality improves, and the camera can be placed somewhere more sensible. On a small stand, for example, slightly above eye level. That is what Continuity Camera does: it lets the Mac use the iPhone camera as its webcam. ## The First Small Obstacle Trying this out is slightly less obvious than you might expect. Most instructions suggest opening a video call in FaceTime or another meeting app. The moment you do that, the iPhone can appear as a camera option. But that approach requires calling someone, which is not always convenient. Not everyone is keen to participate in your technical experiments. The small trick I discovered is that you can simply open **QuickTime Player** , which is installed on every Mac, and start a _New Movie Recording_. In the camera menu you can select the iPhone as the camera. That immediately gives you a live video feed. No call required. It turns out to be the perfect little laboratory: you can move the phone, change the angle, check lighting, and compare it to the built-in Mac camera. Start Quick Time Player and head for the menu bar, New Movie Recording. ## What Actually Changes Two things become apparent quite quickly. First, the image quality difference is real. The iPhone camera produces a noticeably sharper and more stable image than older Mac webcams. Second, the position of the camera matters more than the resolution. When the camera sits slightly above the screen instead of inside it, the framing simply feels more natural. There are also some software features layered on top of this. Center Stage, when supported by the camera, keeps the subject centred by automatically adjusting the frame. And Desk View uses a wide-angle camera and some clever perspective correction to show what is lying on the desk below the screen. Desk View is perhaps the most curious of the three. Without moving the phone, the Mac can generate a second camera view that looks as if there is an overhead camera pointing down at the desk. It is not perfect, but it is surprisingly usable. Interestingly, that trick works with any sufficiently wide-angle webcam, including the ones built into newer iMacs. Desk View ## A Small Upgrade Trick What I came away with is that these features are less about novelty and more about incremental upgrades. If your Mac already has a good camera, the difference may be modest. But if parts of your setup are slightly dated, Continuity Camera is a simple way to upgrade them without buying additional equipment. You are essentially borrowing the camera from the device that already has the best one. ## The Minor Trade-offs There are also a few practical downsides. Using the iPhone as a camera usually means it needs to sit somewhere fixed. A small stand or mount helps. If you travel a lot, that can make the setup a little less spontaneous. And of course there is the obvious one: while the phone is serving as your camera, you cannot casually pick it up and start using it. There is also a small practical dependency: the iPhone now becomes part of your video setup. That means relying on its battery life, or occasionally plugging it in during longer calls or recordings. Still, for something that is already built into the devices, it is a surprisingly capable little toolset. Sometimes the most interesting part of new hardware is simply taking the time to explore the corners you normally ignore. * * * A lighter computer for a different eraA new iMac, a one-hour migration, fewer ports, more cloud and AI. Replacing a desktop computer now reveals how much the role of the machine itself has changed.Rob HoeijmakersRob Hoeijmakers

But a surprising amount of modern work is not limited by knowledge, tools, or intelligence. It is limited by identity. The small rituals of logging in, confirming codes, verifying accounts, and proving, **again and again** , that you are you. ## Quick takeaways * Many digital systems assume that **only the account holder can act**. * Real life, however, runs on **delegation** : partners, colleagues, assistants, children, caregivers. * Two-factor authentication and identity checks make delegation difficult or impossible. * The result is a hidden productivity drain: **identity management work**. * The same constraint also limits what AI agents can actually do. ## The delegation gap In ordinary life, people constantly act on behalf of others. A partner renews insurance. A child handles paperwork for parents. An assistant arranges travel or contracts. A colleague submits documents. This is not exceptional behaviour. It is how daily life functions. Yet most digital systems assume something else entirely: one account, one person, one operator. Security systems reinforce this assumption. Passwords, two-factor authentication, biometrics, device verification. Each action requires the account holder to appear and confirm themselves. This works well for security. But it quietly breaks something else: **delegation**. ## Identity work Because of this design, a surprising part of modern life is spent performing small identity rituals. Logging in. Entering codes. Approving notifications. Verifying devices. These steps are individually trivial. Collectively they form a kind of invisible labour. You might call it **identity work**. It is the administrative layer that sits between intention and action. ## The AI paradox This constraint also explains something curious about AI. AI systems are increasingly capable of planning, analysing, and organising tasks. In principle they could handle many everyday administrative jobs. But they usually cannot. Not because they lack intelligence. Because they lack **authorised identity**. The real systems where action happens, _banks, government portals, insurers, utilities_ , are locked behind personal authentication. So the human remains the final operator, approving and executing steps that machines could otherwise handle. ## When delegation becomes administration Some systems attempt to support delegation through formal mandates or authorisations. But these often turn delegation itself into another administrative task: registering permissions, managing expiry dates, renewing access codes. The effort required to delegate can become almost as large as the task being delegated. ## A missing piece of digital infrastructure What seems largely absent from today’s digital world is a simple capability: **safe, temporary, and limited delegation of digital authority.** The ability to say: * this person may manage this account for a period * these actions are allowed, others are not * access can be revoked at any time Corporate IT systems have such mechanisms. Everyday digital services rarely do. ## Closing reflection For decades we have focused on making information easier to access and process. But productivity may increasingly depend on something more mundane: How easily we can **act on behalf of one another** in digital systems. Real life is cooperative. The internet, in many places, still assumes **we operate alone**. While everyone debates hybrid working, many firms are missing the real productivity killer. It’s not where people work. It’s the mental load they bring with them. Is your team working or are they… | Elvis Eckardt RecruitmentWhile everyone debates hybrid working, many firms are missing the real productivity killer. It’s not where people work. It’s the mental load they bring with them. Is your team working or are they quietly “life-adminning” on your payroll? New workplace data suggests the average UK professional now spends up to 90 minutes of the workday dealing with personal admin. Energy suppliers. GP appointments. Insurance renewals. School logistics. The modern workday is increasingly interrupted by the second shift of life management. And in 2026, some forward-thinking UK SMEs are testing a radical fix: Employer-sponsored Life-Admin Support. Instead of another wellbeing webinar or free fruit basket, companies are partnering with concierge services that handle everyday personal admin for employees. Why? Because burnout is increasingly driven by technostress and constant cognitive overload, not just workload. The results from early adopters are interesting: Measurable increases in deep work time, fewer mid-day distractions, lower burnout among mid-career managers In other words, a focus strategy. The most valuable employee benefit might not be a pay rise that gets swallowed by fiscal drag. It might simply be time and mental space. If companies don’t help employees manage their lives they’ll keep doing it during the workday anyway. Curious where people stand on this: Business owners: Would you trade the annual summer party budget for a service that handles your team’s life admin for a year? Employees: What would you value more, a 2% pay rise or a life concierge that gives you your evenings back?LinkedInElvis Eckardt Recruitment

Er duikt regelmatig een opvallende statistiek op in onderzoek naar werk en productiviteit: werknemers besteden gemiddeld **ongeveer 90 minuten per dag** aan hun eigen administratie. Belastingzaken. Afspraken regelen. Verzekeringen aanpassen. Formulieren invullen. Accounts herstellen. Vaak gebeurt dat gewoon tussendoor, tijdens werktijd. Voor werkgevers lijkt dat in eerste instantie een productiviteitsprobleem. Maar het kan ook een aanwijzing zijn dat er iets anders speelt: dat onze digitale systemen verrassend veel frictie bevatten. ## Waarom persoonlijke administratie zo veel tijd kost Veel administratieve taken zijn inhoudelijk helemaal niet zo ingewikkeld. Het gaat meestal om relatief eenvoudige handelingen: een formulier invullen, een wijziging doorgeven, een afspraak maken. Toch voelen deze processen vaak complex en tijdrovend. Dat komt omdat digitale administratie zelden uit één handeling bestaat. Het is meestal een reeks stappen: * de juiste website vinden * door verschillende schermen navigeren * documenten verzamelen * informatie invoeren * foutmeldingen oplossen En bijna altijd speelt er nog iets anders mee: **digitale identiteit**. ## Eén account, één persoon Vrijwel alle digitale diensten zijn gebouwd rond hetzelfde uitgangspunt: **één account – één persoon – één operator.** Daarom bestaan er wachtwoorden, sms-codes, two-factor authentication, biometrische verificatie en apparaatcontroles. Vanuit veiligheid is dat begrijpelijk. Systemen moeten zeker weten dat degene die iets doet ook daadwerkelijk de eigenaar van het account is. Maar dat ontwerp heeft ook een keerzijde. In het dagelijks leven handelen mensen voortdurend namens elkaar. Partners helpen elkaar met administratie. Kinderen regelen zaken voor hun ouders. Vrienden of collega’s helpen met formulieren. Online werkt dat vaak minder soepel. ## Delegatie wordt administratie Waar iemand in het dagelijks leven simpelweg kan zeggen: “Kun jij dit even regelen?”, ontstaat online al snel een reeks extra stappen. Mandaten. Autorisaties. Nieuwe accounts. Verificatiecodes. Delegatie verandert zo gemakkelijk in een extra administratief proces. Dat helpt misschien verklaren waarom veel mensen hun persoonlijke administratie juist op het werk doen. Daar zitten ze toch al achter een computer, hebben ze documenten bij de hand en kunnen ze stap voor stap door het proces gaan. Het zegt dus niet alleen iets over gedrag, maar ook over hoe digitale systemen zijn ontworpen. ## Een ontwerpprobleem van het internet Wat hier zichtbaar wordt, is een kleine maar fundamentele spanning. Het internet is sterk ontworpen rond **individuele identiteit**. Het dagelijks leven functioneert juist vaak via **samenwerking en delegatie**. Die twee logica’s botsen regelmatig. En zolang digitale systemen zo sterk gekoppeld blijven aan individuele accounts en persoonlijke verificatie, blijft administratie moeilijk te vereenvoudigen of uit te besteden. Niet alleen voor mensen. Maar ook voor technologie. ## Een interessante vraag voor de toekomst Dit maakt persoonlijke administratie op werktijd misschien minder verrassend dan het lijkt. Het is niet alleen een kwestie van discipline of tijdsmanagement. Het kan ook een signaal zijn dat onze digitale infrastructuur nog niet goed aansluit op hoe mensen hun leven organiseren. Waar frictie zit, ontstaan vaak nieuwe ideeën en oplossingen. Maar voordat die oplossingen er zijn, is het misschien eerst de moeite waard om het onderliggende probleem goed te begrijpen: hoe digitale identiteit en delegatie zich tot elkaar verhouden. Daar zou weleens een belangrijk ontwerpthema voor de komende jaren kunnen liggen.

I work daily with large language models, and with ChatGPT in particular. It has become part of how I think, draft, compare perspectives, and test ideas. But there is a side effect: it accumulates. Conversations remain visible. Threads multiply. The system keeps a trail of exploratory thinking that was never meant to be permanent. Recently, I decided to approach that accumulation more deliberately. Not as a technical problem, but as a matter of habit. ## Quick takeaways * Data accumulation has psychological effects, even when storage is cheap. * ChatGPT offers limited middle ground between “keep everything” and “delete everything”. * Periodic, deliberate clearing works better for me than continuous trimming. * Cognitive hygiene is part of responsible AI use. * Personal practice shapes how we later think about organisational data governance. ## How ChatGPT handles conversations In practical terms, the options are relatively simple. Conversations are stored by default. You can archive them, but that mainly changes their visibility. They still exist in the account environment. There is also a more radical option: deleting chats, or deleting all chats. That removes them from the interface and reduces what remains available for continuity across threads. In the Plus environment, there is some cross-thread awareness and temporary memory. If you clear everything, you also lose that contextual continuity. In the Business environment, memory behaves differently, but the structural choice is similar: retain broadly, or remove decisively. There is no refined “delete everything older than three months” button. No nuanced retention schedule. It is either accumulation by default, or active intervention. That simplicity forces a decision. ## The psychology My instinct to save everything did not come out of nowhere. It was shaped in an earlier phase of the internet. There was a time when storage was limited. Hard drives were small. Backups were manual. An export could genuinely save your work, your correspondence, your history. Losing data meant real loss. So you learned to preserve. To copy. To archive. To keep local control. That reflex made sense then. It was protective. The environment has changed completely. Storage is abundant. Cloud systems retain by default. Platforms are designed to accumulate. Conversations, drafts, metadata, versions, everything remains somewhere. Paradoxically, scarcity trained us to save carefully. Abundance now tempts us to save indiscriminately. I noticed that I was still operating with the old reflex in a new context. Keeping everything no longer protected me from loss. It simply increased volume. Letting go, in an era of automatic retention, feels counterintuitive. Deleting something that is effortlessly stored can feel almost irresponsible. And yet, for me, it has become the more disciplined act. Not because data is dangerous in itself, but because unlimited retention creates a diffuse environment. The value of a conversation often lies in what it enabled, not in its permanent storage. Once that became clear, the old reflex began to loosen. ## Hygiene and rhythm I experimented with gradual pruning. It kept the topic alive in my mind. There was always something to clean up. What works better for me is a tidal approach. A period of active use. Then a deliberate session of consolidation and clearing. Before deleting, I sometimes export or summarise what still matters. Often, I realise that very little truly needs to be retained. The core ideas have already moved into notes, documents, or articles. Pressing delete is slightly uncomfortable. There is always a moment of doubt. But the effect afterwards is consistent: the interface feels lighter. The tool feels more like a workspace and less like an archive. ## The business implication This is not only personal preference. In professional contexts, we discuss AI governance, retention policies, compliance, and risk. Those discussions are necessary. But they remain abstract if not grounded in practice. If we treat conversational AI as a permanent cognitive archive, we implicitly accept indefinite retention of exploratory thinking. If we treat it as a workspace, we introduce rhythm, boundaries, and completion. For me, the choice to clean out regularly is aligned with how I now treat data more broadly. Keep what has structural value. Remove what has served its purpose. That stance affects how I think about organisational data as well. Stewardship begins with habit. ## Closing I no longer aim to preserve every intermediate step of thinking. ChatGPT is a powerful amplifier. But amplification does not require permanence. Periodic reset is not a rejection of the tool. It is a way of keeping it usable, proportionate, and aligned with how I want to work. ChatGPT Memory Has Quietly Changed (Plus & Pro)If you use ChatGPT Plus or Pro: memory now goes beyond saved preferences. Past conversations can be retrieved when relevant. Business accounts are different.Rob HoeijmakersRob HoeijmakersAnki, Human Memory in a Digital WorldIn this post, I discuss the importance of human memory in the digital age, where information flows constantly. I also share my workflow for using flashcards and digital tools to retain knowledge.Rob HoeijmakersRob Hoeijmakers

_(en waarom dat voor zakelijke gebruikers logisch is)_ Ik gebruik ChatGPT intensief in mijn werk. Voor analyse, structurering, redactie en het verkennen van scenario’s. Wat ik lange tijd niet actief deed, was mijn gesprekshistorie beheren. Alles bleef staan. Tot ik besefte: dit is feitelijk een extern werkarchief. En elk extern werkarchief vraagt om bewuste omgang. ## ChatGPT is geen kladblok, maar een denkruimte In ChatGPT ontstaan geen formele documenten, maar denksporen. Daarin zitten vaak: * vroege versies van klantideeën * strategische verkenningen * interne reflecties * scenario’s die nooit extern bedoeld waren Juist dat voorlopige karakter maakt ze gevoelig. Eindproducten zijn afgestemd, geredigeerd en bedoeld voor verspreiding. Denksporen niet. Wie ChatGPT intensief inzet, creëert dus een tweede laag informatie binnen zijn professionele praktijk. En die laag verdient aandacht. ## Onzichtbare accumulatie is het echte risico Het gaat zelden om één specifiek gesprek. Het risico zit in accumulatie. Na verloop van tijd ontstaat een archief van: * context * iteraties * half uitgewerkte gedachten * strategische richtingen Niet omdat er iets misgaat. Maar omdat alles automatisch bewaard blijft. Voor zakelijke gebruikers is dat relevant. Informatiebeheer gaat niet alleen over documenten en e-mail. Het gaat ook over denkprocessen die via externe systemen lopen. Als professional wil je globaal weten: * Wat staat er nog? * Hoe ver terug? * In welke context is iets ontstaan? Die overzichtelijkheid is onderdeel van professioneel werken. ## Actief beheren is geen paniekmaatregel Het verwijderen van geschiedenis maakt niets absoluut veilig. Wat eerder verwerkt is, kan onderdeel zijn van systeemprocessen. Dat besef hoort erbij. Maar je verkleint wél: * de hoeveelheid herleidbare context in je account * de cumulatie van gevoelige trajecten * de kans dat oude denksporen opnieuw opduiken Dat is geen wantrouwen. Het is proportioneel handelen. Net zoals je oude back-ups niet onbeperkt bewaart. Of vertrouwelijke concepten niet eindeloos in je downloadsmap laat staan. Digitale hygiëne is geen dramatische ingreep. Het is onderhoud. ## Dit past binnen professioneel datagebruik Voor organisaties is het logisch om na te denken over: * welk type informatie via AI-tools mag lopen * hoe lang interacties bewaard blijven * wie daar toegang toe heeft * hoe dit past binnen bredere governance Voor individuele professionals geldt hetzelfde, zij het op kleinere schaal. Je kunt: * periodiek oude trajecten verwijderen * gevoelige gesprekken afronden en wissen * bewust tijdelijke werkruimtes gebruiken Dat verandert niets aan je verantwoordelijkheid voor inhoud. Maar het voorkomt dat digitale werksporen zich onbeperkt opstapelen. ## Een onverwacht neveneffect Er is nog een tweede effect. Als gesprekken tijdelijk zijn, werk je scherper. Je vat beter samen. Je structureert explicieter. Je archiveert zelf wat blijvend waardevol is. De tool wordt dan geen opslagplaats, maar een werkruimte. En dat is misschien wel de kern. Niet maximale controle. Maar bewuste omgang met digitale denksporen. Voor zakelijke gebruikers is dat geen overdreven voorzichtigheid. Het is professioneel handelen in een tijd waarin steeds meer denkwerk via externe systemen verloopt.

## Inleiding Onlangs las ik dat in een wereld waarin kunstmatige intelligentie steeds meer kenniswerk overneemt, drie menselijke vermogens doorslaggevend blijven: visie, oordeel en creativiteit. Dat klinkt abstract. Maar voor ondernemers en bestuurders is het allesbehalve dat. AI is geen hype meer. Het zit al in software, processen, klantcontact en besluitvorming. De vraag is dus niet óf het impact heeft, maar hoe je als organisatie regie houdt. ## De valkuil van analyse Veel organisaties blijven hangen in analyse. Welke tools zijn er? Wat doen concurrenten? Wat zijn de risico’s? Wat zegt de wetgeving? Dat is begrijpelijk. Het speelveld beweegt snel. Maar analyse alleen leidt zelden tot richting. Sterker nog: het kan verlammend werken. Wie AI uitsluitend benadert als technologische ontwikkeling, loopt het risico om reactief te blijven. Dan wordt de toekomst iets dat je overkomt. ## De verschuiving: van voorspellen naar positioneren De doorbraak zit in een andere vraag. Niet alleen: _Wat gaat AI met ons doen?_ Maar vooral: _Hoe willen wij AI inzetten in onze organisatie?_ Dat is een fundamenteel verschil. AI-capaciteiten zullen toenemen. Marktdruk zal groeien. Efficiëntievoordelen zullen zichtbaar worden. Dat zijn randvoorwaarden. Maar hoe je processen inricht, waar je menselijke controle bewaart, welke verantwoordelijkheden je expliciet maakt, dat zijn keuzes. Daar begint visie. Visie is geen glazen bol. Het is een expliciete voorkeur voor een bepaalde manier van werken, organiseren en waarde creëren. Oordeel gaat over grenzen. Wanneer vertrouwen we een systeem? Wanneer niet? Waar blijft menselijke eindverantwoordelijkheid ononderhandelbaar? Creativiteit gaat over ontwerp. Hoe bouwen we processen waarin AI versterkt zonder te verschralen? Hoe herontwerpen we rollen zodat mensen zich richten op interpretatie, relatie en afweging? ## Vijf jaar vooruit Kijk je vijf jaar vooruit, dan is een aantal zaken waarschijnlijk: * AI is diep geïntegreerd in standaardsoftware. * Kenniswerk wordt deels geautomatiseerd, deels verschoven. * Klanten verwachten snellere, beter onderbouwde antwoorden. * Toezicht en regelgeving worden scherper. De strategische vraag is dan niet of je AI gebruikt, maar hoe. Word je een organisatie die tools stapelt? Of een organisatie die bewust ontwerpt? ## Individuele en collectieve verantwoordelijkheid Voor ondernemers betekent dit: * Investeer niet alleen in tools, maar in beoordelingsvermogen. * Maak expliciet welke waarden je wilt beschermen. * Ontwikkel scenario’s in plaats van één voorspelling. Voor teams betekent het: * Begrijpen wat AI wel en niet kan. * Leren werken mét systemen zonder kritisch denken uit te schakelen. * Transparantie organiseren over gebruik en verantwoordelijkheid. ## Waar dit mij brengt Ik heb zelf gemerkt hoe gemakkelijk het is om te blijven analyseren. Trends volgen. Rapporten lezen. Scenario’s doordenken. Wat veranderde, was de vraag. Niet langer alleen: _Waar gaat dit heen?_ Maar: _Hoe willen wij ons positioneren in dit speelveld?_ Dat is geen afgerond antwoord. Het is een beginpunt. Visie, oordeel en creativiteit zijn geen zachte termen. Ze zijn strategische vermogens. In een tijdperk van intelligente systemen is regie geen vanzelfsprekendheid. Het is een keuze.

<h2 id="notes-from-the-edge-of-building-with-llms">Notes from the edge of building with LLMs</h2><p>I am all in on large language models as cognitive amplifiers. I use them to think, to structure, to explore, to write. That part feels natural to me. What I have never been particularly drawn to is using them as coding tools. Programming has always felt like a different craft. Adjacent, interesting, but not where I naturally place my attention.</p><p>And yet, over the past months, that boundary has become harder to ignore. Not because I suddenly wanted to become a developer, but because clients started asking different questions. They see headlines about “AI eating code”. They see demos racing past on LinkedIn. And they wonder whether they are missing something important.</p><p>So I decided to explore. Not to reskill, but to orient myself. To understand what is actually changing, where agency now sits, and what is realistically accessible to people who are curious, but not intent on becoming software engineers.</p><p>What follows is not a survey. It is a short field note, guided by my own steps.</p><h2 id="standing-near-the-edge-of-coding">Standing near the edge of coding</h2><p>I have never really been a coder. I know enough to follow along, to collaborate, to get something running from <a href="https://github.com/rhoeijmakers">GitHub</a> if I have to, but it has always felt like a different craft. Interesting, sometimes impressive, but not where I naturally operate. You may recognise that position. Curious, not allergic to technology, but not inclined to disappear into documentation or frameworks either.</p><p>What surprised me in these experiments was not how powerful the tools were, but how quietly the boundary shifted. Not because I learned a new skill, but because the effort moved. An idea could turn into something that actually works. Not a mock-up, not a diagram, but a usable thing.</p><p>That was new to me. Not spectacular, but disorienting in a subtle way.</p><h2 id="first-encounter-describing-a-thing-into-existence">First encounter: describing a thing into existence</h2><p>The first thing I tried was <a href="https://lovable.dev">Lovable</a>.</p><p>Someone <a href="https://www.linkedin.com/posts/cassombroek_bij-de-meeste-bedrijven-zijn-e-mailhandtekeningen-activity-7418204193548562432-Gv62?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAAAAGPccBqY4Rvlcj2zZ2PdvOk7tl5dQrkUs">shared a prompt</a> for building a simple email signature generator. I copied it, adjusted it slightly, and watched the result appear. A working interface. Inputs, preview, export. No setup, no explicit architectural decisions on my side.</p><p>What struck me was how familiar this felt. This was not programming in the traditional sense. It felt much closer to product definition. Describing behaviour, constraints, and outcomes. The kind of thinking that normally stops at wireframes or tickets.</p><p>Here, the description executed.</p><p>That is the strength of tools like this. They let you stay at the level of intent and outcome. At the same time, the trade-off is clear. Many decisions are made for you. Hosting, structure, deployment, all of that is implicit. You gain speed and clarity, but give up control.</p><p>For an experiment, that was exactly right.</p><figure class="kg-card kg-bookmark-card"><a class="kg-bookmark-container" href="https://brand-signature-scribe.lovable.app/"><div class="kg-bookmark-content"><div class="kg-bookmark-title">Lovable App</div><div class="kg-bookmark-description">Lovable Generated Project</div><div class="kg-bookmark-metadata"><img class="kg-bookmark-icon" src="https://hoeijmakers.net/content/images/icon/favicon-40.ico" alt="" /><span class="kg-bookmark-publisher">Lovable</span></div></div><div class="kg-bookmark-thumbnail"><img src="https://hoeijmakers.net/content/images/thumbnail/id-preview-7f5b0dcc--bb3425a8-ad34-420a-81f6-0c4ac8723ba9.lovable.app-1768640349762.png" alt="" /></div></a></figure><h2 id="a-step-deeper-intent-meets-a-project">A step deeper: intent meets a project</h2><p>Curiosity pulled me one layer down. I wanted to see what would happen if I stayed close to intent, but entered a recognisable software project.</p><p>I tried <a href="https://www.anthropic.com/engineering/claude-code-best-practices">Claude Code</a> and connected it to GitHub. This time, the result was not an abstract app, but a repository. Files, versions, commits. I used it to build a bespoke email signature generator inside Google Workspace, backed by a Google Sheet and implemented with Google Apps Script.</p><p>The experience changed in kind.</p><p>Claude Code did not “build an app for me”. It helped me think through a project. What goes where. What changes when I adjust something. How to iterate, deploy, and improve. The LLM was no longer replacing structure. It was inhabiting it with me.</p><p>I also felt a boundary. Systematic software development, long-term maintenance, deeper architectural decisions, that still felt like a different craft. Not inaccessible, but distinct.</p><p>What mattered was that this boundary was now visible.</p><figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://hoeijmakers.net/content/images/2026/01/image.png" class="kg-image" alt="" loading="lazy" width="2000" height="1364" srcset="https://hoeijmakers.net/content/images/size/w600/2026/01/image.png 600w, https://hoeijmakers.net/content/images/size/w1000/2026/01/image.png 1000w, https://hoeijmakers.net/content/images/size/w1600/2026/01/image.png 1600w, https://hoeijmakers.net/content/images/size/w2400/2026/01/image.png 2400w" /><figcaption><span style="white-space:pre-wrap">The signature app in Google Workspace.</span></figcaption></figure><h2 id="what-people-mean-when-they-say-%E2%80%9Cide%E2%80%9D">What people mean when they say “IDE”</h2><p>This is where the term “IDE” often comes in. IDE stands for Integrated Development Environment, but that label tends to obscure more than it explains.</p><p>In practice, an IDE is the place where software lives while it is being made and maintained. It is where files are organised, changes are tracked, versions are compared, and decisions accumulate over time. Not just a text editor, but a working environment with memory.</p><p>A widely known example is <a href="https://code.visualstudio.com">Visual Studio Code</a>, used by many developers and familiar to a broader technical audience. What is changing now is not the existence of such environments, but their role.</p><p>With LLMs embedded, IDEs start to absorb more of the thinking, explanation, and iteration that used to happen outside them. They begin to look less like specialist tools and more like shared workspaces between human and system.</p><h2 id="three-ways-of-building-loosely-speaking">Three ways of building, loosely speaking</h2><p>Looking back, what helped me make sense of this was not taxonomy for its own sake, but orientation. Roughly speaking, I now see three modes people move between.</p><p>First, <strong>outcome-driven tools like Lovable</strong>, where you stay at the level of intent and receive a working artefact.</p><p>Second, <strong>artifact-centric assistance</strong>, where LLMs help you produce scripts, components, or configurations that you then integrate yourself.</p><p>Third, <strong>IDE-centred environments</strong>, where the LLM is embedded into the development workflow itself and develops a persistent understanding of a system over time. Tools such as <a href="https://cursor.com">Cursor</a> are often mentioned in this context, as are <a href="https://openai.com/codex/">Codex</a>-based systems from OpenAI.</p><p>These are not strict categories. They overlap and evolve. But they differ in one crucial respect: where your agency sits. At intent. At artefacts. Or at systems.</p><p>What has changed is that moving between these layers no longer requires a hard identity shift.</p><h2 id="a-signal-worth-paying-attention-to">A signal worth paying attention to</h2><p>This is also why it is telling to see Google presenting tools like <a href="https://antigravity.google">Antigravity</a> explicitly as a “next-generation IDE”. Even if such tools are still emerging, the framing itself matters.</p><p>It suggests that IDEs are no longer seen only as specialist developer tools, but as environments where complex work takes shape, especially when that work results in living systems rather than finished documents.</p><p>That matters even if you never plan to work there yourself.</p><figure class="kg-card kg-image-card"><img src="https://hoeijmakers.net/content/images/2026/01/image-1.png" class="kg-image" alt="" loading="lazy" width="2000" height="1159" srcset="https://hoeijmakers.net/content/images/size/w600/2026/01/image-1.png 600w, https://hoeijmakers.net/content/images/size/w1000/2026/01/image-1.png 1000w, https://hoeijmakers.net/content/images/size/w1600/2026/01/image-1.png 1600w, https://hoeijmakers.net/content/images/size/w2400/2026/01/image-1.png 2400w" /></figure><h2 id="abundance-noise-and-judgement">Abundance, noise, and judgement</h2><p>One effect of this shift deserves a little more attention. As software becomes easier to create and easier to discard, the market around it changes shape.</p><p>We move away from a relatively structured landscape, with high entry costs and clear thresholds, towards something more crowded and informal. A place with many small offerings, many claims, many tools competing for attention. At times, it starts to resemble a medina: lively, inventive, and full of possibility, but also noisy and disorienting.</p><p>In such an environment, the hard part is no longer building something. The hard part is knowing what needs to be built, what can be thrown away, and what deserves care, continuity, and quality.</p><p>Not every problem needs a robust system. Not every tool should live beyond its moment.</p><figure class="kg-card kg-image-card kg-card-hascaption"><img src="https://hoeijmakers.net/content/images/2026/01/30626148-E457-4C58-836E-EEAE655E76B5_1_105_c.jpeg" class="kg-image" alt="" loading="lazy" width="1024" height="768" srcset="https://hoeijmakers.net/content/images/size/w600/2026/01/30626148-E457-4C58-836E-EEAE655E76B5_1_105_c.jpeg 600w, https://hoeijmakers.net/content/images/size/w1000/2026/01/30626148-E457-4C58-836E-EEAE655E76B5_1_105_c.jpeg 1000w, https://hoeijmakers.net/content/images/2026/01/30626148-E457-4C58-836E-EEAE655E76B5_1_105_c.jpeg 1024w" /><figcaption><span style="white-space:pre-wrap">Finding your way in buying or using software is changing. </span></figcaption></figure><h2 id="what-actually-gets-democratised">What actually gets democratised</h2><p>Software development is becoming more democratic. Not because coding no longer matters, but because much of the handwork no longer determines who gets to participate.</p><p>It is now relatively easy to spin something up. To test an idea. To build a situational tool and discard it again. Software starts to behave more like a medium than a monument.</p><p>At the same time, this does not remove the hard parts. Vision still matters. Knowing what to build, for whom, and why does not get automated away. Nor does the work of building systems that are robust, scalable, secure, and efficient.</p><p>If anything, that work becomes more valuable.</p><p>Good developers do not become less important here. They become more important. Not because they type faster, but because they understand structure, trade-offs, and long-term consequences. The same goes for people who know how to deploy systems into production, test them properly, and keep them running once they stop being experiments.</p><p>What changes is the division of labour. A large amount of manual work has quietly been absorbed by tools. That opens space for non-coders to explore, and for coders to focus on what actually requires judgement.</p><p>So when people say “you no longer need coders”, they are pointing at the wrong thing. What is disappearing is not the craft, but the gatekeeping.</p><p>And in a noisier, more abundant landscape, that kind of discernment becomes the real advantage.</p><hr /><figure class="kg-card kg-bookmark-card"><a class="kg-bookmark-container" href="https://hoeijmakers.net/excel-and-the-future-cockpit-of-business-logic/"><div class="kg-bookmark-content"><div class="kg-bookmark-title">Excel and the future cockpit of business logic</div><div class="kg-bookmark-description">Excel has always been more than a spreadsheet. For decades it has been the place where business logic quietly lives. Not in software systems designed for control, but in the free space where analysts, planners and managers actually think. What interests me today is how this space is changing as</div><div class="kg-bookmark-metadata"><img class="kg-bookmark-icon" src="https://hoeijmakers.net/content/images/icon/Circle-logo-2-503.png" alt="" /><span class="kg-bookmark-author">Rob Hoeijmakers</span><span class="kg-bookmark-publisher">Rob Hoeijmakers</span></div></div><div class="kg-bookmark-thumbnail"><img src="https://hoeijmakers.net/content/images/thumbnail/IMG_5794-2.jpeg" alt="" /></div></a></figure><figure class="kg-card kg-bookmark-card"><a class="kg-bookmark-container" href="https://www.forbes.com/sites/joemckendrick/2026/01/15/smaller-nimbler-smarter-ai-taking-paths-of-least-resistance/"><div class="kg-bookmark-content"><div class="kg-bookmark-title">Smaller, Nimbler, Smarter: AI Taking Paths Of Least Resistance</div><div class="kg-bookmark-description">With AI projects this year, there will be less of a push to boil the ocean, and instead more of a laser-like focus on smaller, more manageable projects.</div><div class="kg-bookmark-metadata"><img class="kg-bookmark-icon" src="https://hoeijmakers.net/content/images/icon/144X144-F.png" alt="" /><span class="kg-bookmark-author">Forbes</span><span class="kg-bookmark-publisher">Joe McKendrick</span></div></div><div class="kg-bookmark-thumbnail"><img src="https://hoeijmakers.net/content/images/thumbnail/0x0.jpg" alt="" /></div></a></figure><figure class="kg-card kg-bookmark-card"><a class="kg-bookmark-container" href="https://hoeijmakers.net/excel-hidden-operating-system-of-business-reasoning/"><div class="kg-bookmark-content"><div class="kg-bookmark-title">Excel, the Hidden Operating System of Business Reasoning</div><div class="kg-bookmark-description">Excel has long been the silent operating system of business reasoning. AI may be about to extend that logic into natural language.</div><div class="kg-bookmark-metadata"><img class="kg-bookmark-icon" src="https://hoeijmakers.net/content/images/icon/Circle-logo-2-504.png" alt="" /><span class="kg-bookmark-author">Rob Hoeijmakers</span><span class="kg-bookmark-publisher">Rob Hoeijmakers</span></div></div><div class="kg-bookmark-thumbnail"><img src="https://hoeijmakers.net/content/images/thumbnail/IMG_8930-1.jpeg" alt="" /></div></a></figure>

<p><br /><strong>Waarom dit idee nu relevant wordt voor groeiende bedrijven in de EU</strong></p><h2 id="kernboodschap">Kernboodschap</h2><p>De Europese Unie werkt aan het idee van een nieuwe, optionele bedrijfsvorm die vanaf de start grensoverschrijdend inzetbaar is. Niet om nationale BV’s, GmbH’s of SARL’s te vervangen, maar om een alternatief te bieden voor bedrijven die <strong>EU-breed willen opschalen zonder telkens opnieuw het juridische wiel uit te vinden</strong>.</p><p>Voor ondernemers gaat dit niet over oprichten alleen, maar over wat daarna komt: groeien, mensen aannemen in meerdere landen, aandelenstructuren opzetten en kapitaal aantrekken binnen Europa.</p><h2 id="wat-is-eu-inc-precies">Wat is EU-INC precies?</h2><p>EU-INC is de informele naam voor een voorgestelde <strong>pan-Europese rechtsvorm</strong>, ook wel het “28ste regime” genoemd. Het idee is dat bedrijven kunnen kiezen voor één Europese juridische structuur die in alle lidstaten wordt erkend.</p><p>Belangrijke uitgangspunten:</p><ul><li>EU-INC is <strong>optioneel</strong>. Nationale rechtsvormen blijven bestaan.</li><li>Belastingen en arbeidsrecht blijven <strong>nationaal geregeld</strong>.</li><li>De focus ligt op <strong>vennootschapsrecht, governance en kapitaalstructuren</strong>.</li></ul><p>Je kunt het zien als een extra juridische laag die bedoeld is om grensoverschrijdend ondernemerschap eenvoudiger en consistenter te maken, zonder de nationale autonomie volledig los te laten.</p><figure class="kg-card kg-bookmark-card"><a class="kg-bookmark-container" href="https://www.eu-inc.org/"><div class="kg-bookmark-content"><div class="kg-bookmark-title">EU–INC — One Europe. One Standard. — Pan-European legal entity.</div><div class="kg-bookmark-description">EU–INC is a proposal for a pan-European standardized legal entity to unlock pan-European startup scaling.</div><div class="kg-bookmark-metadata"><img class="kg-bookmark-icon" src="https://www.chatvoorbedrijven.nl/content/images/icon/hTcgaWRewUNZeB6wVrfBSXk9Mn0.svg" alt="" /></div></div><div class="kg-bookmark-thumbnail"><img src="https://www.chatvoorbedrijven.nl/content/images/thumbnail/DywwAp236AdnECpow4p9VcGluY.png" alt="" /></div></a></figure><h2 id="waarom-dit-voorstel-nu-weer-op-tafel-ligt">Waarom dit voorstel nu weer op tafel ligt</h2><p>In veel EU-landen is het oprichten van een bedrijf inmiddels snel en digitaal geregeld. Dat is niet het knelpunt.</p><p>De complexiteit ontstaat zodra een bedrijf:</p><ul><li>actief wordt in meerdere landen</li><li>personeel aanneemt buiten het land van oprichting</li><li>aandelen, opties of converteerbare leningen wil uitgeven</li><li>investeerders aantrekt uit andere EU-lidstaten</li></ul><p>Elke extra jurisdictie betekent een nieuw juridisch systeem, met eigen regels voor bestuur, aandeelhouders, rapportage en compliance. Dat maakt groei duurder, trager en risicovoller. Voor investeerders is dit vaak een reden om voorkeur te geven aan bedrijven die buiten Europa zijn gestructureerd.</p><p>EU-INC probeert precies dát probleem te adresseren.</p><h2 id="wat-is-de-status-nu">Wat is de status nu?</h2><p>Er is <strong>nog geen wet</strong> en er is <strong>nog geen definitieve juridische structuur</strong>. Wat er wel is:</p><ul><li>Een duidelijke politieke oproep aan de Europese Commissie om met een concreet voorstel te komen.</li><li>Toenemende steun vanuit startup-ecosystemen, investeerders en beleidsmakers.</li></ul><p>Als de Commissie met een voorstel komt, volgt daarna een langdurig traject van onderhandelingen met lidstaten. Het gaat dus om jaren, niet om maanden.</p><p>Voor ondernemers verandert er vandaag niets. Maar de <strong>richting is duidelijker dan in eerdere pogingen</strong>.</p><h2 id="waarom-eerdere-pogingen-strandden">Waarom eerdere pogingen strandden</h2><p>Het idee van één Europese bedrijfsvorm is niet nieuw. Eerdere voorstellen, zoals de European Private Company, zijn nooit breed ingevoerd.</p><p>Terugkerende bezwaren zijn onder andere:</p><ul><li>nationale controle over vennootschapsrecht</li><li>angst voor juridische ‘shoppen’ tussen systemen</li><li>zorgen over arbeidsrechten en sociale bescherming</li><li>de vraag of bestaande nationale structuren niet voldoende zijn</li></ul><p>Ook de naam “EU-INC” roept weerstand op. Voor sommigen klinkt die te Amerikaans of te sterk gericht op aandeelhouderswaarde.</p><h2 id="wat-zou-eu-inc-concreet-kunnen-betekenen-voor-bedrijven">Wat zou EU-INC concreet kunnen betekenen voor bedrijven?</h2><p>Als het voorstel er daadwerkelijk komt, zou EU-INC in theorie bieden:</p><ul><li>één juridische entiteit geldig in de hele EU</li><li>één set governance-regels</li><li>eenvoudiger uitgifte van aandelen en opties</li><li>een centraal Europees register in plaats van nationale versnippering</li></ul><p>Daarmee kan het voor groeiende bedrijven makkelijker worden om EU-breed te opereren zonder telkens juridische herstructureringen.</p><p>Tegelijk blijft het belangrijk om realistisch te blijven:</p><p>belasting, sociale zekerheid en arbeidsrecht blijven nationaal. EU-INC haalt frictie weg, maar elimineert die niet volledig.</p><h2 id="wat-heb-je-hier-als-ondernemer-n%C3%BA-aan">Wat heb je hier als ondernemer nú aan?</h2><p>EU-INC is geen instrument dat je morgen kunt gebruiken. Het is wel relevant om:</p><ul><li>te begrijpen <strong>waar Europa economisch naartoe wil</strong></li><li>te volgen als je ambities of investeerders EU-breed zijn</li><li>mee te nemen in strategische discussies over vestigingsstructuren</li></ul><p>Het voorstel raakt aan een grotere vraag: wil Europa een interne markt zijn waar kapitaal en ondernemerschap écht kunnen schalen, of blijft groei vooral nationaal georganiseerd?</p><h2 id="afsluitend">Afsluitend</h2><p>EU-INC is geen revolutie op korte termijn. Het is een signaal dat Europa zoekt naar manieren om ondernemerschap, innovatie en kapitaalvorming minder versnipperd te maken.</p><p>Voor ondernemers en adviseurs is het vooral iets om <strong>te volgen, te begrijpen en te plaatsen</strong>. Niet omdat het morgen alles verandert, maar omdat het veel zegt over hoe de Europese markt zich de komende tien jaar zou kunnen ontwikkelen.</p><figure class="kg-card kg-bookmark-card"><a class="kg-bookmark-container" href="https://ec.europa.eu/commission/presscorner/detail/en/speech_26_150"><div class="kg-bookmark-content"><div class="kg-bookmark-title">Special Address by President von der Leyen at the World Economic Forum</div><div class="kg-bookmark-description">It is now 55 years since the first meeting here in Davos. The idea of the founder, Klaus Schwab, was to create a platform to discuss the issues and the ideas of the day. Of course, the world has transformed completely since 1971. But the original idea of Davos has remained, as we have just heard in the speeches. So I was delighted that you have gone back to your roots with this year\’s theme – A Spirit of Dialogue.</div><div class="kg-bookmark-metadata"><img class="kg-bookmark-icon" src="https://static.ghost.org/v5.0.0/images/link-icon.svg" alt="" /><span class="kg-bookmark-author">European Commission - European Commission</span></div></div><div class="kg-bookmark-thumbnail"><img src="https://www.chatvoorbedrijven.nl/content/images/thumbnail/sm_ec_logo_big.jpg" alt="" /></div></a></figure>

When the EU says it is building eight quantum computers, the number sounds precise. It isn’t. This piece explores why that question is harder than it looks.

WhatsApp is voor veel bedrijven vooral een communicatiemiddel. Met Channels krijgt het ook publieke reach. Dat verklaart waarom de EU nu meekijkt.

WhatsApp is introducing usernames. A small change that reshapes identity, privacy, and how people and businesses can be reached without sharing phone numbers.

Met usernames laat WhatsApp het telefoonnummer los. Voor gebruikers betekent dat minder delen, voor bedrijven krijgt WhatsApp eindelijk een voorkant.

We talk about unlimited data, but most of our digital life quietly depends on fibre. What mobile, roaming and holidays reveal about real limits.