
Congrats on the PASAP launch, @deepenai! Excited to help make AV safety more measurable together. 🤝

Founder & CEO @deepenai Working on AI to make our lives safer & more productive. ex-Google, Munch.vc

Congrats on the PASAP launch, @deepenai! Excited to help make AV safety more measurable together. 🤝

Today's featured data on Deepen Refinery. (Link in comments)

We started Deepen AI in 2017 with a mission: help build autonomous systems that save lives and make their safety measurable. PASAP brings years of that work together with @nvidia Halos SAFE. Grateful to everyone who helped us get here. deepen.ai/press-releases/dee…

Https://services.deepen.ai Mechanical turk has shut down! And this is our answer to offer an alternative


6/ Join us at refinery.deepen.ai

5/ For data owners: list on your terms, block named competitors, stay anonymous through beta, get paid per sale. For builders: evaluate the sample, the license, and the validation before you commit a dollar.

4/ Yesterday we opened the waitlist for Deepen Refinery: a marketplace for real-world Physical AI data. Egocentric video, LiDAR, multimodal. Every dataset is scored, deduplicated, and rights-cleared before a buyer sees it.

3/ That is the data the next generation of AI is being built on. Nobody can judge a dataset before they buy it. Nobody who owns good data has a safe way to sell it. So builders re-collect and owners sit on drives. Both lose.

2/ Authentic real-world data, with provenance and the rights to train on it. Look inside the largest public training sets and the audits find: ~30% duplicate images in LAION-2B 70%+ license omission on popular dataset hosts 6%+ wrong labels in ImageNet validation

1/Most of what we built at @deepenai went into ADAS and self-driving. Tens of teams got safer programs on the road with our calibration, labeling, and validation tools. The next generation is robots, humanoids, and world models. They need something automotive never had to solve.

Physical AI doesn’t just need more data. It needs refined data. Validated. Calibrated. Traceable. Built with provenance. That’s why we built Deepen Refinery: a marketplace for real-world data you can actually trust. refinery.deepen.ai/

The design bet: self-improving agents don't fail because they can't write skills, they fail because they confidently install junk. Evidence + a human gate fixes that. Free & MIT-licensed: github.com/mmusa/second-natu…

Three commands: 🌱 /harvest: mines your sessions for repeated work, drafts it into skills with evidence attached ✅ /promote: nothing self-installs; every skill passes a human gate ✂️ /curate: weekly pass that retires what's stale (archives, never deletes)

Self-improving agents are the hot idea, but today that means running a standalone daemon: separate memory, separate skills, a security surface you now own. I wanted the compounding without the second brain. So I built Second Nature, native in @claudeai

Customers are looking for outcomes and outcomes need credible experts with strong judgement to stand behind it linkedin.com/posts/mohammad_…