ai•Lab @ailab.ee · 28/07/2026The models you choose make a big difference. Do your homework before making any long-term commitments. 020
ai•Lab @ailab.ee · 24/07/2026Don’t just blindly automate your company’s work with whatever LLM you’re used to. This graph is a perfect illustration of why it’s important to pick the right model for every task. The cost difference over the years will be huge. 020
ai•Lab @ailab.ee · 22/07/2026Everything in Publisher's UI is now a REST API. Articles, AI drafting, scheduling, social, approvals, analytics, knowledge base, connections. Org-scoped keys intersect a member's permissions. UI and API stay in lockstep. ailab.ee/2026/06/new-how-publisher-…...ailab.eeThe new Publisher release puts your analytics behind every draft and schedule – AI LabPublisher has changed from a tool for writing and scheduling into a data-driven content engine. The platform now tracks real-world performance across your accounts, feeds those insights directly into an 020
ai•Lab @ailab.ee · 16/07/2026Mixture-of-experts is the cheapest path to high intelligence per dollar. Only the relevant expert sub-network fires per token. The model behaves like a smaller specialist on every call. ailab.ee/2026/06/intelligence-per-d…ailab.eeIntelligence Per Dollar Is the New Buyer Question – AI LabMany companies are blowing through their budgets on hungry AI tools that burn cash with wordy, inefficient answers. To keep costs under control, businesses must stop looking at flat pricing 020
ai•Lab @ailab.ee · 15/07/2026The Commission's high-risk AI classification guidelines are open for stakeholder feedback. If your AI is on the boundary, this is where you push back. Details: ailab.ee/2026/06/is-your-ai-system-…ailab.eeIs Your AI System High-Risk? The EU’s New Draft Guidelines, Explained – AI LabOn 19 May 2026 the European Commission published draft guidelines on how to classify high-risk AI systems under Article 6 of the AI Act. Consultation is open until 23 June 110
ai•Lab @ailab.ee · 14/07/202612-16 extra months on high-risk AI compliance. Three things to do with the runway: 1. Recheck classification 2. Build, not defer 3. Verify SMC threshold ailab.ee/?p=1093ailab.eeWhat the EU AI Omnibus Just Changed for AI Builders – AI LabThe AI Omnibus provisional agreement has now cleared Coreper and Parliament committee approval, with formal adoption still pending. For AI builders, the working deal delays high-risk AI obligations, gives SMEs 010
ai•Lab @ailab.ee · 13/07/2026Small software teams are going to look bigger. Not because agents remove engineers, but because each engineer can supervise more useful work: research, code, tests, docs, migrations. The advantage shifts to fast judgment. ailab.ee/2026/06/the-next-ai-develo…ailab.eeThe Next AI Development Bottleneck Is Judgment – AI LabAnthropic's recursive self-improvement essay is most useful as an engineering signal, not a doomsday forecast. Its internal data suggests code writing is becoming less scarce, while specification, review, testing, and 020
ai•Lab @ailab.ee · 09/07/2026Three levers cut AI cost-per-outcome in production. Pick the right size model. Measure tokens at the application layer. Route cheap-by-default with premium escalation. Stop reading the pricing page. ailab.ee/2026/06/intelligence-per-d…ailab.eeIntelligence Per Dollar Is the New Buyer Question – AI LabMany companies are blowing through their budgets on hungry AI tools that burn cash with wordy, inefficient answers. To keep costs under control, businesses must stop looking at flat pricing 010
ai•Lab @ailab.ee · 08/07/2026Article drafting in Publisher is now run end to end by an autonomous agent. Decides what the piece needs. Researches only when called for. Pulls from your knowledge base. Close the tab, come back to a finished draft. ailab.ee/2026/06/new-how-publisher-…...ailab.eeThe new Publisher release puts your analytics behind every draft and schedule – AI LabPublisher has changed from a tool for writing and scheduling into a data-driven content engine. The platform now tracks real-world performance across your accounts, feeds those insights directly into an 000
ai•Lab @ailab.ee · 08/07/2026Privacy in Play Text is the simplest kind. Your text never leaves your device. No server for it to go to. No telemetry. No accounts. Voices generated on-device. ailab.ee/2026/06/we-built-a-free-te…ailab.eeWe built a free text-to-speech app in two evenings, and here is how it works – AI LabWe built Play Text, a free and private text-to-speech app for Mac and iOS, in just two evenings. It turns articles, PDFs, and web pages into audio with live, word-by-word 100
ai•Lab @ailab.ee · 07/07/2026Centralised enforcement at the EU AI Office is the quietly important part of the AI Omnibus deal. Less member-state fragmentation, more consistent reads. Full take: ailab.ee/?p=1093ailab.eeWhat the EU AI Omnibus Just Changed for AI Builders – AI LabThe AI Omnibus provisional agreement has now cleared Coreper and Parliament committee approval, with formal adoption still pending. For AI builders, the working deal delays high-risk AI obligations, gives SMEs 000
ai•Lab @ailab.ee · 06/07/2026The best first use cases for coding agents are not giant rewrites. Start with regression fixes, tests, dependency upgrades, docs drift, migrations, and narrow refactors with clear checks. Agents need rails before freedom. ailab.ee/2026/06/the-next-ai-develo…ailab.eeThe Next AI Development Bottleneck Is Judgment – AI LabAnthropic's recursive self-improvement essay is most useful as an engineering signal, not a doomsday forecast. Its internal data suggests code writing is becoming less scarce, while specification, review, testing, and 020
ai•Lab @ailab.ee · 03/07/2026WordPress sync in the new Publisher: fix a typo in Publisher, it lands on the live WordPress post in place. No new draft, no -2 duplicate. Reschedules and retries update the existing post. Self-heals stale references. ailab.ee/2026/06/whats-new-in-publi…...ailab.eeWhat’s New in Publisher: An MCP Server, Automatic Pipelines, and a WordPress Overhaul – AI LabPublisher's latest update introduces an MCP server that lets AI assistants securely draft, edit, and publish content directly inside your workspace. The release also adds automated content pipelines and a 000
ai•Lab @ailab.ee · 02/07/2026Seven sectors where your AI defaults to high-risk under EU rules: - Education - Employment - Essential services - Law enforcement - Justice - Biometrics - Critical infra ailab.ee/2026/06/is-your-ai-system-…ailab.eeIs Your AI System High-Risk? The EU’s New Draft Guidelines, Explained – AI LabOn 19 May 2026 the European Commission published draft guidelines on how to classify high-risk AI systems under Article 6 of the AI Act. Consultation is open until 23 June 000
ai•Lab @ailab.ee · 01/07/2026Play Text ships with Kokoro-82M baked in. 327 MB of neural voice, running locally on Apple silicon via MLX. Natural-sounding speech. No API key, no cloud bill, works offline. ailab.ee/2026/06/we-built-a-free-te…ailab.eeWe built a free text-to-speech app in two evenings, and here is how it works – AI LabWe built Play Text, a free and private text-to-speech app for Mac and iOS, in just two evenings. It turns articles, PDFs, and web pages into audio with live, word-by-word 020
ai•Lab @ailab.ee · 30/06/2026Microsoft's new coding model uses up to 60% fewer tokens than Claude Haiku 4.5. The trick is adaptive solution length. Short answers for easy tasks, longer ones only when needed. ailab.ee/2026/06/intelligence-per-d…ailab.eeIntelligence Per Dollar Is the New Buyer Question – AI LabMany companies are blowing through their budgets on hungry AI tools that burn cash with wordy, inefficient answers. To keep costs under control, businesses must stop looking at flat pricing 210
ai•Lab @ailab.ee · 29/06/2026Lines of code are becoming a worse productivity metric. Anthropic says its engineers merge about 8x more code per day than in 2024. The real signal is pressure moving to review, tests, architecture boundaries, and release judgment. ailab.ee/2026/06/the-next-ai-develo…...ailab.eeThe Next AI Development Bottleneck Is Judgment – AI LabAnthropic's recursive self-improvement essay is most useful as an engineering signal, not a doomsday forecast. Its internal data suggests code writing is becoming less scarce, while specification, review, testing, and 020
ai•Lab @ailab.ee · 26/06/2026Publisher pipeline: drop a link in the queue, it drafts on-brand, asks for approval, publishes to WordPress, stitches the final URL into social copy automatically, and lands them in your schedule's next slots. ailab.ee/2026/06/whats-new-in-publi…...ailab.eeWhat’s New in Publisher: An MCP Server, Automatic Pipelines, and a WordPress Overhaul – AI LabPublisher's latest update introduces an MCP server that lets AI assistants securely draft, edit, and publish content directly inside your workspace. The release also adds automated content pipelines and a 000
ai•Lab @ailab.ee · 25/06/2026If your firm is under 750 people and €150M revenue, the AI Act just got simpler. SMC compliance relief is real. What changed and why it matters: ailab.ee/?p=1093ailab.eeWhat the EU AI Omnibus Just Changed for AI Builders – AI LabThe AI Omnibus provisional agreement has now cleared Coreper and Parliament committee approval, with formal adoption still pending. For AI builders, the working deal delays high-risk AI obligations, gives SMEs 000
ai•Lab @ailab.ee · 24/06/2026Select text in any Mac app, Mail, Slack, a PDF, anywhere. Press the Play Text shortcut. A floating player opens with the full text and starts reading. Always-on-top so you can move on to the next thing. ailab.ee/2026/06/we-built-a-free-te…...ailab.eeWe built a free text-to-speech app in two evenings, and here is how it works – AI LabWe built Play Text, a free and private text-to-speech app for Mac and iOS, in just two evenings. It turns articles, PDFs, and web pages into audio with live, word-by-word 000
ai•Lab @ailab.ee · 23/06/2026The AI in Publisher now consults your real numbers before drafting. Your top posts. Best channels. Winning formats. Strongest timing. The agent shapes each draft to match what works for your audience. ailab.ee/2026/06/new-how-publisher-…ailab.eeThe new Publisher release puts your analytics behind every draft and schedule – AI LabPublisher has changed from a tool for writing and scheduling into a data-driven content engine. The platform now tracks real-world performance across your accounts, feeds those insights directly into an 010
ai•Lab @ailab.ee · 23/06/2026A weak ticket now creates a large vague diff. A strong specification creates a controlled agent task. As coding agents take on longer work, specs become more valuable than tickets. Clear goals, constraints, and acceptance checks beat better prompts. ailab.ee/2026/06/the-next-ai-develop...ailab.eeThe Next AI Development Bottleneck Is Judgment – AI LabAnthropic's recursive self-improvement essay is most useful as an engineering signal, not a doomsday forecast. Its internal data suggests code writing is becoming less scarce, while specification, review, testing, and 231
ai•Lab @ailab.ee · 22/06/2026Publisher shipped today. It now reads your real performance data and feeds it back into the AI that drafts and schedules. A content tool that gets measurably smarter from the posts you publish. ailab.ee/2026/06/new-how-publisher-…ailab.eeThe new Publisher release puts your analytics behind every draft and schedule – AI LabPublisher has changed from a tool for writing and scheduling into a data-driven content engine. The platform now tracks real-world performance across your accounts, feeds those insights directly into an 010
ai•Lab @ailab.ee · 22/06/2026The industry has a name for AI waste now: tokenmaxxing. Amazon staff were caught burning tokens to climb internal leaderboards. Models pad answers to game benchmarks. Both versions hit the same invoice. ailab.ee/2026/06/intelligence-per-d…ailab.eeIntelligence Per Dollar Is the New Buyer Question – AI LabMany companies are blowing through their budgets on hungry AI tools that burn cash with wordy, inefficient answers. To keep costs under control, businesses must stop looking at flat pricing 110
ai•Lab @ailab.ee · 19/06/2026Tuesday morning workflow with the new Publisher MCP server: tell Claude to draft a post from your notes, schedule it on WordPress, queue 4 social variants. It happens inside Publisher. Approvals still required. Token-scoped. ailab.ee/2026/06/whats-new-in-publi…...ailab.eeWhat’s New in Publisher: An MCP Server, Automatic Pipelines, and a WordPress Overhaul – AI LabPublisher's latest update introduces an MCP server that lets AI assistants securely draft, edit, and publish content directly inside your workspace. The release also adds automated content pipelines and a 110
ai•Lab @ailab.ee · 18/06/2026Two ways your AI gets classified high-risk under Article 6: 1. Safety component of an Annex I regulated product 2. Use case from Annex III Full breakdown: ailab.ee/2026/06/is-your-ai-system-…ailab.eeIs Your AI System High-Risk? The EU’s New Draft Guidelines, Explained – AI LabOn 19 May 2026 the European Commission published draft guidelines on how to classify high-risk AI systems under Article 6 of the AI Act. Consultation is open until 23 June 000
ai•Lab @ailab.ee · 17/06/2026Open a long Safari article. Click Play Text. Choose "Read article" or "Pick a section". A pink pill sweeps word by word on the live page as it speaks. You stay in context, you just stop reading. ailab.ee/2026/06/we-built-a-free-te…ailab.eeWe built a free text-to-speech app in two evenings, and here is how it works – AI LabWe built Play Text, a free and private text-to-speech app for Mac and iOS, in just two evenings. It turns articles, PDFs, and web pages into audio with live, word-by-word 010
ai•Lab @ailab.ee · 16/06/2026This week's LLM charts, explained simply. Three different leaders: Claude Fable 5 is smartest, MiniMax M3 is now most-used on OpenRouter, and Claude is hardest to trick (F5). No single model wins all three. ailab.ee/2026/06/llm-charts-explain…ailab.eeLLM charts explained, week of 16 June 2026 – AI LabThe first post in a weekly, plain-language read of the AI model leaderboards. This week: a tight intelligence race led by Claude Fable 5, MiniMax M3 surging to the top 010
ai•Lab @ailab.ee · 16/06/2026The thing about the new Publisher release: MCP server, pipelines, and WordPress rebuild all share the same publishing code underneath. Agent, human, or pipeline, the result is the same. Mixed workflows finally feel safe. ailab.ee/2026/06/whats-new-in-publi…...ailab.eeWhat’s New in Publisher: An MCP Server, Automatic Pipelines, and a WordPress Overhaul – AI LabPublisher's latest update introduces an MCP server that lets AI assistants securely draft, edit, and publish content directly inside your workspace. The release also adds automated content pipelines and a 000
ai•Lab @ailab.ee · 15/06/2026Publisher just shipped its biggest update yet. 58 changes in two weeks. Highlights: MCP server (Claude works inside your workspace), end-to-end pipelines, and a rebuilt WordPress integration with two-way sync. ailab.ee/2026/06/whats-new-in-publi…...ailab.eeWhat’s New in Publisher: An MCP Server, Automatic Pipelines, and a WordPress Overhaul – AI LabPublisher's latest update introduces an MCP server that lets AI assistants securely draft, edit, and publish content directly inside your workspace. The release also adds automated content pipelines and a 021
ai•Lab @ailab.ee · 12/06/2026EU just published the missing manual on classifying high-risk AI. The Commission's draft guidelines walk through Article 6 in plain terms. ailab.ee/2026/06/is-your-ai-system-…ailab.eeIs Your AI System High-Risk? The EU’s New Draft Guidelines, Explained – AI LabOn 19 May 2026 the European Commission published draft guidelines on how to classify high-risk AI systems under Article 6 of the AI Act. Consultation is open until 23 June 010
ai•Lab @ailab.ee · 12/06/2026Behind Play Text. Our founder wanted to listen to articles like a podcast. Existing apps cost too much for what they did. Two evenings with Claude Code on Opus 4.8 later, Mac + iOS apps existed. Free. ailab.ee/we-built-a-free-text-to-sp…ailab.eeWe built a free text-to-speech app in two evenings, and here is how it works – AI LabWe built Play Text, a free and private text-to-speech app for Mac and iOS, in just two evenings. It turns articles, PDFs, and web pages into audio with live, word-by-word 120
ai•Lab @ailab.ee · 11/06/2026We just shipped Play Text. A free text-to-speech app for Mac and iPhone with live word-by-word highlighting. Built in two evenings. Free. No cloud, no subscription, no tracking. ailab.ee/we-built-a-free-text-to-sp…ailab.eeWe built a free text-to-speech app in two evenings, and here is how it works – AI LabWe built Play Text, a free and private text-to-speech app for Mac and iOS, in just two evenings. It turns articles, PDFs, and web pages into audio with live, word-by-word 710
ai•Lab @ailab.ee · 10/06/2026The next AI development bottleneck is not writing code. It is judgment. Anthropic says 80%+ of code merged into its own codebase in May was Claude-authored. If agents produce code faster than teams can review it, review becomes the system. ailab.ee/the-next-ai-development-bo…...ailab.eeThe Next AI Development Bottleneck Is Judgment – AI LabAnthropic's recursive self-improvement essay is most useful as an engineering signal, not a doomsday forecast. Its internal data suggests code writing is becoming less scarce, while specification, review, testing, and 000
ai•Lab @ailab.ee · 09/06/2026Anthropic just released Claude Fable 5, making a Mythos-class model safe for general use. It hits state-of-the-art on nearly all tested benchmarks and beats their past public models. We are testing if it justifies the commotion. www.anthropic.com/news/claude-fable…anthropic.comClaude Fable 5 and Claude Mythos 5Today we’re launching Claude Fable 5: a Mythos-class model that we’ve made safe for general use. 000
ai•Lab @ailab.ee · 09/06/2026GPT-5.5 and Claude Opus 4.8 score within a point on the Intelligence Index. Running it costs $3,357 on one, $4,685 on the other. Same answer, 40% more expensive. ailab.ee/intelligence-per-dollar-is…ailab.eeIntelligence Per Dollar Is the New Buyer Question – AI LabMany companies are blowing through their budgets on hungry AI tools that burn cash with wordy, inefficient answers. To keep costs under control, businesses must stop looking at flat pricing 100
ai•Lab @ailab.ee · 08/06/2026Microsoft cancelled Claude Code licenses across its biggest engineering division. Two weeks later they shipped their own coding model that uses 60% fewer tokens. The two announcements are the same story. ailab.ee/intelligence-per-dollar-is…ailab.eeIntelligence Per Dollar Is the New Buyer Question – AI LabMany companies are blowing through their budgets on hungry AI tools that burn cash with wordy, inefficient answers. To keep costs under control, businesses must stop looking at flat pricing 010
ai•Lab @ailab.ee · 08/06/2026You finally have the headroom to test less-technical AI agents. Claude Cowork just doubled its 5-hour usage limits for Pro, Max, and Team users. Time to get hands-on and learn what actually works for you. support.claude.com/en/articles/1540…support.claude.comClaude Cowork June 2026 usage promotion | Claude Help Center 010
ai•Lab @ailab.ee · 05/06/2026EU just delayed high-risk AI obligations. Annex III to 2 Dec 2027, Annex I to 2 Aug 2028. The new timeline and what to do with the extra 12-16 months: ailab.ee/what-the-eu-...ailab.eeWhat the EU AI Omnibus Just Changed for AI Builders – AI LabOn 7 May 2026 the Council of the EU and the European Parliament reached a provisional agreement on the AI Omnibus. High-risk AI obligations are pushed out by 12 to 011
ai•Lab @ailab.ee · 26/03/2026Needed a tool to track office listings. Our AI assistant built one in minutes. Needed competitor monitoring? Built that too. We gave it the ability to create custom micro apps on the fly. ailab.ee/our-ai-executive-assistant…ailab.eeOur AI Executive Assistant Creates Purpose-Built Apps on the Fly – AI LabWe help businesses implement AI solutions that actually work. From strategy to execution, we turn AI potential into measurable results. 000
ai•Lab @ailab.ee · 21/03/2026We gave our AI agent its own M365 account. It reads emails, detects promises, and once apologized on our CEO's behalf while he was stuck in traffic. Works ~80% of the time. The other 20% keeps us humble. ailab.ee/giving-an-ai-agent-its-own…ailab.eeGiving an AI Agent Its Own Microsoft 365 Account – AI LabWe help businesses implement AI solutions that actually work. From strategy to execution, we turn AI potential into measurable results. 000
ai•Lab @ailab.ee · 19/03/2026What does it take to give an AI agent its own email? Dedicated sub-agents, a communication etiquette system, and a lot of trial and error. Here's what we built and what broke. ailab.ee/giving-an-ai-agent-its-own…ailab.eeGiving an AI Agent Its Own Microsoft 365 Account – AI LabWe help businesses implement AI solutions that actually work. From strategy to execution, we turn AI potential into measurable results. 000
ai•Lab @ailab.ee · 18/03/2026Together with @heysec.com we built a fully autonomous cybersecurity research factory. It researches prompt injection vulnerabilities, builds research cases with PoC exploits, creates firewall rules to mitigate these threats, and iteratively tests those against active exploits. 010
ai•Lab @ailab.ee · 16/03/2026We quietly released a new product a week ago. Originally developed as an internal AI agent, it has since evolved into a full platform with a visual studio and advanced agent orchestration. ailab.ee/support-agentailab.eeSupport Agent – AI LabWe help businesses implement AI solutions that actually work. From strategy to execution, we turn AI potential into measurable results. 011
ai•Lab @ailab.ee · 12/03/2026These agents' confusion moments are a great example of this other 20-30%. 👉 ailab.ee/we-turned-openclaw-into-a-… 011
ai•Lab @ailab.ee · 11/03/2026We’re putting the OpenClaw EA to the real test and asking it to handle the move of our offices. Let’s see how that goes 🤞 👉 We published more about what this AI EA really is all about: ailab.ee/we-turned-openclaw-into-a-… 010
ai•Lab @ailab.ee · 10/03/2026You can now also listen to all our blog posts 🎧 Give it a try on the latest blog post about how we built an EA on OpenClaw: ailab.ee/we-turned-openclaw-into-a-… 110
ai•Lab @ailab.ee · 09/03/2026We turned OpenClaw into an AI Executive Assistant that handles real email, calendar, and communications every day. It works about 70-80% of the time. The other 20-30% has been... educational. ailab.ee/we-turned-openclaw-into-a-…ailab.eeWe Turned OpenClaw Into a Fully Functioning Executive Assistant – AI LabWe help businesses implement AI solutions that actually work. From strategy to execution, we turn AI potential into measurable results. 010
Reposted by ai•LabKlemens Arro @klemens.bsky.social · 22/02/2026We’ve now built agents that autonomously optimize and improve other teams of agents. This includes optimizing not only data or prompts but the entire codebase. Of course, it includes guardrails and easy rollbacks to maintain stability and recover from possible mistakes. 022