Amplio
DeepMind's harness for autonomous research runs that last days to weeks. Used internally on Simply, now public.
Feature a build or sponsor the directory. DM @pranavmore69
6 hand-picked research projects shipped on X. Every card is a real demo or repo, not a screenshot thread.
DeepMind's harness for autonomous research runs that last days to weeks. Used internally on Simply, now public.
Harness that automates 'keep going' on open math problems.
Can a VLM act through a body? Open-source benchmark for action intelligence, decoupled from motor control.
A method for rational inquiry that pushes a question right to the evidential limit and no further, to avoid both credulity and conspiracy.
Four official Anthropic plugins (Financial Analysis, Market Researcher, Equity Researcher, Earnings Reviewer) for deep stock research.
Shows real paywalls and onboarding flows of apps making $40k/month so you can study and rebuild them.
Glad that others are finding Amplio helpful. It is our main harness for autonomous long research runs spanning days to weeks. It is heavily used by our team for automated AI research / RSI on top of the Simply codebase. Open sourced at: github.com/google-deepmin…
Harness matters. We test Gemini 3.1 Pro with amplio, a harness for robust and long-horizon runs and super fit for auto-research. We then get much better results than with Gemini CLI. Harness: github.com/google-deepmin… FrontierCS is an open-ended benchmark for Computer Science
Introducing Autoprover. AI’s newest math meme: give a model an open problem and keep replying “keep going.” Sometimes it actually works - so I built a harness that automates that flow. Just on steroids. github.com/guyz/autoprover Show more
🔥 HumanCLAW is now fully open-sourced! github.com/Human-CLAW/Hum…
🦾HumanCLAW: Can Vision-Language Models Act Through a Body? A VLM can spot the sofa in a second. But can it get a body there and sit down? We call it Action Intelligence, and our work decouples it from motor control so it can finally be measured. Today’s best models turn out to
AI can be too credulous. Or too conspiratorial. The trick is pushing inquiry right to the evidential limit — but no further. I built a tool for that. The “coffin corner” method of rational inquiry Now try to break it: newsletter.martingeddes.com/p/the-coffin-c…
Nearly every retail investor doing stock research is working from data that's already stale by the time they read it. Anthropic just shipped four plugins that close that gap Claude Pro is required for access, the free tier won't unlock the tools Setup runs through the toolbar. Show more
An AI assistant just got its own phone number and email address. It makes your calls, waits on hold, and books the follow-ups. And it is one of 7 things that dropped this week most people slept on. Me and @AndrewWarner broke them all down in the video below. Here's what you Show more