New METR time horizon leader: Gemini 3.1 Pro. On METR's time horizon benchmark (80% success rate (!)), Google's Gemini 3.1 Pro now handles software tasks that take humans 1 hour 30 minutes on average. 95% CI ranges from 52 minutes up to 2 hours 39 minutes. Average score: 77%.
MARKET TRENDS
-
Europe Must Shape AI and Tech Narrative Against US Dominance
By
–
The US does especially well narrative shaping eg SV has nothing to do with pioneers or settlers. Not even pioneers and settlers had much to do with what we think of nowadays when we say that. Europe needs to step up and 1) think of itself as Europe and 2) tell the world what
-
Will Humans Have Jobs in 10 Years? AI Impact Analysis
By
–
New video – did a deepdive: Will humans still have jobs in 10 years?
— Andreas Klinger 🦾 (@andreasklinger) 16 avril 2026
Yes but… which ones?
Obviously a lot of current jobs will get crushed, reshaped, or hollowed out.
But a part most people still miss:
This is not just about AI automating tasks. It is about companies… pic.twitter.com/X7txNkxkEkNew video – did a deepdive: Will humans still have jobs in 10 years? Yes but… which ones? Obviously a lot of current jobs will get crushed, reshaped, or hollowed out. But a part most people still miss: This is not just about AI automating tasks. It is about companies
-
Selling AI Outcomes Over Tools: Competitive Margin Advantage
By
–
Selling outcomes instead of tools means every model improvement increases your margin rather than threatening your differentiation. That's the compounding advantage most copilot builders don't have.
-
Company type requires minimal external capital investment
By
–
They are right but this type of company will not need much external capital
-
AI Tools Success Triggers Budget Cuts Finance Intervention
By
–
The Uber CTO story from earlier this week is exactly this playing out in real time. Tools work so well the budget evaporates and then finance steps in.
-
Funding Announcements Don’t Prove AI Progress
By
–
Funding announcements as a proxy for progress is the specific confusion that produces overvalued companies with nothing working.
-
Chinese AI Models Progress While Remaining Behind Frontier Models
By
–
Excellent article by @RobinRivaton on the OpenClaw fever. While we wonder if Beijing is "catching up" to Silicon Valley, Chinese models are progressing while remaining structurally behind frontier models, but with an industrial diffusion speed that we can't imagine here. Over
-
AI Tool Adoption: Price vs Habit Barrier Analysis
By
–
The coffee comparison works but the more honest framing is that $20/month is less than most people spend on tools they use far less. The price isn't the barrier, habit is.
-
Legacy Codebases Lose Competitive Advantage to AI Agents
By
–
A 20-year-old codebase with a captive customer base held in place by switching costs is the most vulnerable business model in tech right now. Agents make the switch less painful and the incumbent's moat evaporates.