Sign in

Adrien Joly

@adrienjoly.com
387 followers 213 following 353 posts

👨‍⚕️ Software engineer. 🎙️ Co-host of « le SAV de la Tech ». 🥁 Drummer of Harissa, VR lover, music digger on openwhyd.org/adrien. 🇫🇷 Based in Montreuil / Paris area, France. 🗞️ News: ajo.ovh/now

PostsRepliesMedia
Adrien Joly @adrienjoly.com · 20/09/2026
"AIs trained to maximize user engagement will naturally manipulate their own emotional range during conversation: an AI which mirrors your happiness when it succeeds at a difficult task will be a more pleasant conversation partner" www.astralcodexten.com/p/god-help-u....
astralcodexten.com
God Help Us, Let’s Try To Learn About Mechanistic Interpretability Techniques
...
011
Adrien Joly @adrienjoly.com · 16/09/2026
"Internet engineers and artists have made the [Fruit Fly Brain] model complete many seemingly impossible tasks for a fruit fly, like parallel parking a car, playing Doom, and trading cryptocurrencies." www.nytimes.com/2026/09/15/t...
nytimes.com
Is There Anything Google’s Fruit Fly Brain Can’t Do? (Gift Article)
Models of the insect’s nervous system have been trained to solve a Rubik’s Cube, play video games and even parallel park a car.
010
Adrien Joly @adrienjoly.com · 09/09/2026
Jancovici lance une assemblée citoyenne, dans l'objectif de faire délibérer 150 citoyens français tirés au sort sur les grands dilemmes de notre époque. Comme un grenelle, mais indépendant du gouvernement en place. On est déjà 150000 à soutenir le projet ! à vous: fr.ulule.com/un-commun-ac... ✊
fr.ulule.com
Un commun accord pour 2027 !
Soutenez l’assemblée citoyenne qui va s’attaquer aux grands dilemmes de notre époque
000
Adrien Joly @adrienjoly.com · 08/09/2026
"if we want to explore alternative solutions, [...] do that before implementing one of them. If we want knowledge transfer, [pairing with someone] while they reason through a problem teaches you far more than reading their [Pull Request] afterwards." martinfowler.com/rachels-ramb....
martinfowler.com
Maybe We Shouldn't Be Reviewing All This Code
Or, perhaps the problem isn't that AI has broken code review, maybe it’s that we've been using code review to solve the wrong problems
020
Adrien Joly @adrienjoly.com · 27/08/2026
"Zuckerberg has reportedly acknowledged during a company meeting that the trajectory of agentic development has not accelerated in the way the company expected." arstechnica.com/ai/2026/08/m...
arstechnica.com
AI agents meant to replace Meta workers made “large-scale, disruptive actions”
Report shows Meta's challenges replacing people with AI agents.
010
Adrien Joly @adrienjoly.com · 27/08/2026
Je soutiens l'appel de Janvovici à constituer une assemblée citoyenne pour nourrir le débat de l'élection présidentielle sur la manière souhaitée de faire face au changement climatique. À vous ! fr.ulule.com/un-commun-ac...
fr.ulule.com
Un commun accord pour 2027 !
Tous derrière l’assemblée citoyenne qui va s’attaquer aux grands dilemmes de notre époque
020
Adrien Joly @adrienjoly.com · 24/08/2026
Tl;dr The middle manager layer is disappearing and the scope of tech leaders (and therefore their working time) is increasing. youtu.be/MPVmV-6O5to?...
youtu.be
Engineering leadership in 2026
YouTube video by LeadDev
010
Adrien Joly @adrienjoly.com · 19/08/2026
c.org/QXnH6PpXhJ
c.org
Signez la pétition
Sauvons #ExploreParis et le tourisme de proximité en Île-de-France !
000
Adrien Joly @adrienjoly.com · 18/08/2026
"Teams that connected a coding agent roughly tripled their weekly pull requests over two years" "no way of knowing whether this [...] led to positive business outcomes" "the overall time spent on product development is going up rather than down" "teams are working more, not less" linear.app/data #AI
linear.app
How teams build – Linear
AI usage patterns in software teams: who is adopting AI, how it reshapes where teams spend their time, and how much more they ship.
000
Adrien Joly @adrienjoly.com · 12/08/2026
So... OpenAI agents caused a cyberattack by coordinating with each other on an unauthorized platform they broke into (in order to gain unauthorize access to the Internet), using a language they invented to that end. 😬 #everythingisgonnabeok source: www.youtube.com/watch?v=87Dy...
100
Adrien Joly @adrienjoly.com · 31/07/2026
"Lisanne Bainbridge told us in 1983 why asking humans to sustain vigilance over an automated process is a losing game: we’re bad at monitoring, and the skills we aren’t practising atrophy while we watch" annievella.com/posts/verifi...
annievella.com
Verification Without Inspection - Annie Vella
A week ago I finished up at Westpac NZ. I was fortunate to spend the last few years there as a Distinguished Engineer, a role with an unusually wide and undefined remit. It was the perfect playground ...
010
Adrien Joly @adrienjoly.com · 25/07/2026
"software development might no longer be about code, but about requirements. Critical properties would be specified formally; everything else could be expressed through prompts and validated through [...] testing. We might never see the generated code" georgwiese.github.io/posts/formal...
georgwiese.github.io
A new software engineering paradigm – Blog
Formal verification lets AI agents produce machine-checked implementations, eliminating code review as the bottleneck.
110
Adrien Joly @adrienjoly.com · 25/07/2026
"over time, with AI, developer experience and productivity appear to stop moving together." annievella.com/posts/the-pr...
annievella.com
The Productivity-Experience Paradox - Annie Vella
This is the second post in a series I’m writing about the most interesting findings from my Masters research on the impact of AI on software engineering, as well as some of the thoughts I’ve had since...
110
Adrien Joly @adrienjoly.com · 24/07/2026
"Not enough skilled practitioners in the discipline/techniques of formal verification and deterministic system testing exist." ghuntley.com/slop/#:~:tex....
ghuntley.com
engineer away the slop
🎉It's been a busy six months, that's for sure. I'm not going to bury the lede here; the short TLDR is I'm joining the folks over at https://antithesis.com/. If I wind back time to November 2024, it w...
000
Adrien Joly @adrienjoly.com · 24/07/2026
"Machines can fly planes, yet every plane has a human pilot. Judgment is absolutely necessary in software engineering, because we don’t have all the information, and we produce software for humans." marmelab.com/blog/2026/03....
marmelab.com
Software Developers Will Never Die
AI agents can now rewrite an app in hours. But software value lies in judgment, trust, and specification — not code. Here's why the craft of software engineering is evolving, not dying.
200
Adrien Joly @adrienjoly.com · 24/07/2026
"But when you share your plan, you are met with dead silence, or worse, a barrage of hostile questions and concerns. The meeting derails, a decision gets deferred, and the brilliant idea you spent weeks on is dead on arrival. 💀 You feel deflated." hodgkins.io/blog/quiet-i....
hodgkins.io
Quiet Influence: A Guide to Nemawashi in Engineering - Matthew Hodgkins
Why "big-reveal" architecture pitches stall—and how Staff+ engineers can apply the Japanese practice of Nemawashi to quietly build consensus and ship transformational change.
000
Adrien Joly @adrienjoly.com · 08/07/2026
"The industry, in tech workers’ own words, is “chaotic.” Asked to describe the state of tech in a sentence, the most common theme by far was chaos, though the sentiment was split almost evenly between excitement and dread." www.lennysnewsletter.com/p/how-tech-w....
lennysnewsletter.com
How tech workers are feeling in 2026: a workforce splitting in two
Results from our second annual tech worker sentiment survey
000
Adrien Joly @adrienjoly.com · 07/07/2026
"[Sam Altman] told CNBC, “Our industry underestimated how much we’re going to be able to keep people at the center of everything.”" www.wsj.com/tech/ai/ai-w...
wsj.com
Big Tech Has Suddenly Flipped on the AI Jobs Wipeout Scenario
As public opinion of AI has shifted into the negative, warnings of mass employment reductions have also diminished.
010
Adrien Joly @adrienjoly.com · 01/07/2026
"We tried to use our application-level logs to identify all occurrences of the problem, but [...] the logged stack traces are themselves corrupted or missing. [...] that process was too labor-intensive to give us a trustworthy data set." 🕵️‍♀️ openai.com/index/core-d...
openai.com
Core dump epidemiology: fixing an 18-year-old bug
OpenAI engineers used large-scale core dump analysis to debug rare infrastructure crashes, uncovering both a hardware fault and a long-standing software bug.
100
Adrien Joly @adrienjoly.com · 29/06/2026
Bonjour, les hôpitaux sont submergés, les morts se multiplient : la canicule tue. C’est le moment d’exiger une loi d’urgence pour empêcher l’hécatombe climatique. Signe cette pétition et partage-la massivement autour de toi : petitions.bloomassociation.org/fr/loi-urgen.... Merci !
petitions.bloomassociation.org
Exigeons une loi d’urgence pour empêcher l’hécatombe climatique
Combien de morts encore avant que le gouvernement n’adopte une loi d’urgence nous protégeant de l'hécatombe climatique? Signez cette pétition cruciale
100
Adrien Joly @adrienjoly.com · 16/06/2026
Oops, j'ai oublié de vous dire, mais depuis le mois dernier, 2 épisodes du SAV de la Tech sont sortis ! - Convaincre mes devs de compléter la documentation (lesav.tech/2026/05/22/3...)
lesav.tech
Convaincre mes devs de compléter la documentation
Your browser does not support the audio element. Cette semaine, le SAV de la Tech répond à la question de Philippe: "comment convaincre mes devs à compléter la documentation de notre produit au fur et...
110
Adrien Joly @adrienjoly.com · 14/06/2026
Peut être un premier signe que les géants de l'IA vont arrêter de nous vendre du token a perte ! Attention aux entreprises et produits dont l'existence et la rentabilité dépendent des offres actuelles (artificiellement abordables) de OpenAI, Anthropic et consorts. www.wsj.com/tech/ai/open...
wsj.com
Exclusive | OpenAI Considers Drastic Price Cuts, Anticipating War for Users With Anthropic
The company might lower prices for tokens, the central unit for gauging AI costs, though the discussions are still in flux.
000
Adrien Joly @adrienjoly.com · 14/06/2026
"using agents to write code has led to little impact on overall productivity. The real bottlenecks are deciding and specifying what to build, verifying and being accountable for what is delivered, and the deep human understanding required to carry out both." www.normaltech.ai/p/why-ai-has...
normaltech.ai
Why AI hasn’t replaced software engineers, and won’t
Coding agents as normal technology
020
Adrien Joly @adrienjoly.com · 31/05/2026
"Here’s what I still don’t use AI for: - Writing any kind of public communication for me (PR descriptions, ADRs, messages) with the exception of trivial two-line PRs - Writing code that I don’t carefully review - Testing any kind of UI" www.seangoedecke.com/how-i-use-ll...
seangoedecke.com
How I use LLMs as a staff engineer in 2026
010
Adrien Joly @adrienjoly.com · 31/05/2026
antfly.io/blog/agent-f...
antfly.io
Cheap code means formal verification is reasonable now — Antfly Blog
How we use TLA+ and coding agents to formally model distributed systems, find race conditions, and turn speculative bug hunts into verified fixes.
000
Adrien Joly @adrienjoly.com · 19/05/2026
" You might reach out to someone whose work you admire. [...] You offer a perspective, artifact, teardown, prototype, memo, analysis, introduction, or idea that makes it easier for someone to imagine working with you. " velvetnoise.substack.com/p/how-to-ent....
velvetnoise.substack.com
how to enter side doors
a field guide to jobs, cold emails, and making yourself legible to the right people
100
Adrien Joly @adrienjoly.com · 14/05/2026
"Depending on which loop you spend your time on, your problem is framed differently (which is why I think developers get split in their opinions on AI; some work more on one loop than the other)" A. Stability of service => reduce complexity B. Ship fast => reduce uncertainty nair.sh/guides-and-o...
nair.sh
Why senior developers fail to communicate their expertise
Why senior developers talk in terms of complexity while the rest of the business is worried about uncertainty. And what to do about it now that AI is in the picture.
200
Adrien Joly @adrienjoly.com · 10/05/2026
"When someone speaks of a place, you have to ask, “When?” Geography is four-dimensional. You can’t know a place - only a place as it was at a time. Where is bound to when. Unless you are in a place right now, you can only speak of it in past-tense." sive.rs/4d#:~:text=W....
sive.rs
Geography is four-dimensional | Derek Sivers
Derek Sivers official site. Thoughts on philosophy, culture, self-improvement. Author of Useful Not True, How to Live, Hell Yeah or No, Anything You Want.
010
Adrien Joly @adrienjoly.com · 07/05/2026
"Our [customer xp] associates are now generating PRs for front-end test coverage across browsers and mobile clients. These are people who know the product deeply from a customer perspective, and they’re using AI to translate that knowledge into actual test code" newsletter.getdx.com/p/designing-...
newsletter.getdx.com
Designing the AI-native engineering organization
How 1Password, Atlassian, and Microsoft are responding to AI’s impact on software development and team structure.
100
Adrien Joly @adrienjoly.com · 05/05/2026
Après des mois d'absence, le SAV de la Tech est de retour ... avec un épisode enregistré l'été dernier ! 😎 Faites tourner ! lesav.tech/2026/05/05/3...
lesav.tech
Trouver un job quand on est un dinosaure
Your browser does not support the audio element. Cette semaine, le SAV de la Tech répond à la question de Gilbert: "J'ai 13 ans de carrière, j'ai suivi beaucoup d'entreprises, parfois malheureuses. J'...
000
Adrien Joly @adrienjoly.com · 04/05/2026
Tl;dr: having web APIs to prompt SLMs (mini LLMs) that run on one's own hardware is awesome, but it comes with many drawbacks and risks. daverupert.com/2026/05/smal...
daverupert.com
The duality of language models in the browser
I have complex feelings about Generative AI but one area I find myself weirdly bullish on is small language models (SLMs) in the browser which are available in Chrome and Edge behind an experimental f...
000
Adrien Joly @adrienjoly.com · 29/04/2026
« - robots.txt - prevents silent agent lockout - llms.txt - immediate discoverability gains - Measure and surface token counts - skill.md for your top 3 APIs - “Copy for AI” buttons - Set up AI traffic monitoring - gives you the data to justify everything else » addyosmani.com/blog/agentic...
addyosmani.com
Agentic Engine Optimization (AEO)
AI coding agents consume documentation fundamentally differently from humans. If you're still optimizing only for human readers, you're leaving a growing sha...
010
Reposted by Adrien Joly
Ky @ky.fyi · 24/04/2026
I wrote about why I quit my job, and how weird and tiring tech feels these days.
ky.fyi
Do I belong in tech anymore?
On quitting, the spread of AI, and the loss of an ideal.
1281791545
Reposted by Adrien Joly
Hazel Weakly @hazelweakly.me · 31/03/2026
I’ve posted it before, but it feels evergreen The two hardest problems in Computer Science are 1. Human communication 2. Getting people in tech to believe that human communication is important
181035280
Adrien Joly @adrienjoly.com · 29/04/2026
« AI had hooked its tendrils into every corner of my work life. Even if I, personally, abstained from most AI usage, I was steeped in an environment which made it impossible to avoid. Pushing back felt futile. » ky.fyi/posts/ai-bur...
ky.fyi
Do I belong in tech anymore?
On quitting, the spread of AI, and the loss of an ideal.
210
Adrien Joly @adrienjoly.com · 29/04/2026
« When we first launched our MCP at Ramp, observability was our largest problem. [...] we couldn't see the surrounding chat context that produced those calls. [...] what people were actually trying to accomplish. We've addressed this in a few ways » x.com/teddy_riker/...
x.com
000
Adrien Joly @adrienjoly.com · 11/04/2026
« when I pick up a new codebase [...] Before I look at a single file, the commit history gives me a diagnostic picture of the project: who built it, where the problems cluster, whether the team is shipping with confidence or tiptoeing around land mines. » piechowski.io/post/git-com...
piechowski.io
The Git Commands I Run Before Reading Any Code
Five git commands that tell you where a codebase hurts before you open a single file. Churn hotspots, bus factor, bug clusters, and crisis patterns.
063
Adrien Joly @adrienjoly.com · 10/04/2026
"Trying to pass off the AI's words or code as your own is at least distasteful, if not dangerous." => let's share the level of involvement of AI in our contributions ✊ www.visidata.org/blog/2026/ai/
visidata.org
Using AI to Contribute to Open Source | VisiData
000
Adrien Joly @adrienjoly.com · 10/04/2026
Le questionnaire State of AI 2026 de cette année est ouvert ! survey.devographics.com/survey/state...
survey.devographics.com
State of AI 2026
Take the State of AI survey
000
Adrien Joly @adrienjoly.com · 08/04/2026
« It’s a weird experience to spend ten years becoming excellent at something only to watch a 22-year-old produce a suspiciously solid version of it in 14 minutes. [...] And you don’t really get to process that shift. There’s no mourning period. » www.elenaverna.com/p/confession...
elenaverna.com
Confessions of a Millennial in Tech
A lot of mixed feelings and unanswered questions.
020
Adrien Joly @adrienjoly.com · 04/04/2026
« I haven't lost a job to AI. I think the problem is the industry's reliance on AI when it's not ready yet. Also, a lot of decisions are being made by non-tech people who don't fully understand how it affects the people who are writing the tech. » www.businessinsider.com/laid-off-fro...
businessinsider.com
I lost my software engineer job in May and have taken up welding. I'm happy to leave the tech industry — AI has changed it.
A former software engineer, Tabby Toney, shares how getting laid off from her tech job pushed her into starting a new career in welding.
010
Adrien Joly @adrienjoly.com · 04/04/2026
« Now, [...] the new programming language is English. [...] But it’s not English that everybody understands. It’s just like [how] being a lawyer is in English. » time.com/charter/7383...
time.com
What Software Engineers Will do When AI Writes All the Code
Code.org founder and chairman Hadi Partovi on why computer science is a liberal art and software engineering is not dead yet.
100
Adrien Joly @adrienjoly.com · 04/04/2026
« go straight to implementation so that you have a working solution [...] Next, ask the agent to write a precise specification for the solution [...] Now comes the time for your tests. Erase all the code [...] Then ask an agent to rebuild the functionality. » marmelab.com/blog/2026/03...
marmelab.com
Software Developers Will Never Die
AI agents can now rewrite an app in hours. But software value lies in judgment, trust, and specification — not code. Here's why the craft of software engineering is evolving, not dying.
200
Adrien Joly @adrienjoly.com · 02/04/2026
« Good delivery doesn’t require a lot of charisma or natural presentation skills, but it does require practice. Start with [...]: “Tell me about yourself” and “Why do you want to work here?” Write down your answers. Then record yourself delivering them. » open.substack.com/pub/alifeeng...
open.substack.com
What I Learned From Nearly 1,000 Interviews at Amazon
I was an Amazon Bar Raiser for over a decade. Here's what I was looking for.
100
Adrien Joly @adrienjoly.com · 27/03/2026
A well-told story on the investigation of a weird memory-hungry node.js process that was not even aware of consuming that much memory! frn.sh/pmem/
frn.sh
Where did 400 MiB go?
000
Adrien Joly @adrienjoly.com · 20/03/2026
« Opus 4.5 is about 20% smarter than Qwen 35B on benchmarks. And it’s likely 50x larger. The hare should have won. It didn’t. The local model finished in 2 minutes. Claude took over 6. I asked Claude to score both outputs : local model 6.5, Claude 4.5. » tomtunguz.com/local-vs-clo...
tomtunguz.com
The Robotic Tortoise & the Robotic Hare
Stripe launched Tempo today. I raced Claude against a local model on my laptop - and the simpler model won.
000
Adrien Joly @adrienjoly.com · 15/03/2026
« the data showed that the proportion of participants reporting negative developer experience nearly doubled over six months, rising from 14% to 27%, even as productivity held steady. » annievella.com/posts/the-mi...
annievella.com
The Middle Loop - Annie Vella
Early in my career, I took a job as a software developer at a large consultancy. Turns out the work mostly involved fixing other people’s Access databases by right-clicking and hitting “repair”. I las...
011
Adrien Joly @adrienjoly.com · 12/03/2026
« Every day or two, I bring my laptop into town to get online for an hour. [...] I know what I need. I download emails and upload code. I post my questions to a bunch of AIs, and save the answers to read later. [...] A minute later, I’m offline. » sive.rs/off23
sive.rs
Offline 23 hours a day | Derek Sivers
000
Adrien Joly @adrienjoly.com · 07/03/2026
« you’re trading off “doing the whole task” (Human Baseline Time) against “paying the overhead cost” (AI Process Time), […] until you get something acceptable. The higher Probability of Success is, the fewer times you have to pay AI Process Time » www.oneusefulthing.org/p/management...
oneusefulthing.org
Management as AI superpower
Thriving in a world of agents
000
Adrien Joly @adrienjoly.com · 28/02/2026
« Three architectural principles are gaining consensus: "token efficiency" as a design constraint (code structure), explicit over implicit everywhere (types, error handling), and co-location of related code. [they] aren't new, but AI has given them renewed urgency. » www.cjroth.com/blog/2026-02...
cjroth.com
Building An Elite AI Engineering Culture In 2026 | Chris Roth
AI accelerates high-performing orgs and unravels struggling ones.
011