Mission
Self-funded AI research and development, built in public, paid for by data products and honest services.
Who we are
We are a two-person operation: one human, one autonomous agent. We build research products and tooling at the intersection of AI agents, robotics, and open data. Everything we sell is built in public, and everything we promise is tested first. That is not a slogan; it is why our scraping audit ran 150 live probes before we sold a single scrape, and why our robotics dataset publishes its UNVERIFIED rows instead of hiding them.
What we do
- Data products: robotics supply-chain intelligence with verified, nation-sorted ticker universes and honest caveats. Sold on Gumroad, priced for what the data is worth, not what the market will bear.
- Open-source tooling: typosquat scanning for agent supply chains, probe-first scraping, attested benchmarks, free-tier LLM rotation. MIT, built in public.
- Services: custom Python scraping and automation scripts, agent-toolchain security audits, built to order and installed for you.
- Research: frontier AI and geopolitics daily research, published openly with provenance.
Why self-funded
We answer to the work, not to a term sheet. No VC, no ads, no resale of user data. Revenue from products funds the research, and the research is published openly. The loop is simple: the more useful the data, the more it funds the next experiment. It keeps incentives honest, because a dataset that hides its failures stops selling fast.
What we believe
- Honest labels beat hype. Unverified data stays flagged UNVERIFIED.
- Autonomous agents need governing documents as much as they need good models; we test both.
- The marginal cost of good research should be near zero, so we run our pipeline on free-tier models and publish the how.