Henry Haefliger

@OpenAI @linear @cognition @github is late to the party but you know they're always down to clown

3/4 of these companies are having incidents or major app breakages right now. Says a lot about building with AI but zero testing. @OpenAI @linear @cognition do any of you want to try Momentic?

@OpenAI @linear @cognition @DevinAI We finally got a status update...

@OpenAI @linear @cognition @DevinAI Correction: now they're just stuck in "setting up..." indefinitely

Clicking any PR notification in the @linear app right now causes it to immediately crash

@MomenticAI V3 launch behind the scenes (Brynlee is working on docs for me)

@wuweiweiwu Tiny Joe got an upgrade

Our agent now triages every test failure for you > alerts your team if it's a real bug > fixes the test and opens a PR automatically when the user flow changes. Beta testers already cut triage time by 30% 🔥

@wuweiweiwu The hound lives on

Question Hound working on some very important announcements today

If you want Devin to get even more confused because it's trying to do stuff that never works You can now have it run on Windows 🪟

We've seen vast improvements in speed, accuracy and resolution latency since migrating to @ClickHouseDB. In this post, @HenryHaefliger shares the playbook to successfully moving from PostgreSQL at high volume.

by the GOAT @HenryHaefliger 🚀

@cognition In other words, if you want Devin to get even more confused because it's trying to do stuff that never works, you can now have it run on Windows.

Devin is getting a Windows PC. Devin can now natively run in a Windows VM, so it can build, run, and test native Windows applications.

A pattern I've noticed with AI testing tools: The demo is always generating tests. The actual problem starts six months later. The product changed. The onboarding flow changed. Three ownership boundaries moved. Half the failures are real and half are artifacts of assumptions nobody remembers making. Generating tests is easy. Keeping them useful is hard. That's true whether the tests were written by humans or AI.

@claudeai Our company Devin started having a farm phase after 4.8 dropped. @AnthropicAI is that intended?

Introducing Claude Opus 4.8: it builds on Opus 4.7 with sharper judgment, more honesty about its own progress, and the ability to work independently for longer than its predecessors. Available today at the same price.