The Index

Companies

793 dossiers · 390 need attention

Two things live here: the startups we follow closely, and the research index of every company we cover. See what we researched this week.

A machined card-index drawer half open, one copper tab standing proud.

The practice / 75 tracked

The startups we follow closely, not just cover

75 read daily23 sectors152 open roles across 20 companies

Run one of these? We do the work first. Then you get a page, a film and a call.For founders

All 75 startups

01SF ComputedatacentersSF Compute sells GPU time by start date and length and lets buyers resell what they don't use, which makes it the closest thing to a spot and forward market for a GPU-hour.Series A
02Agnost AIContinual learningWatches live agent conversations for silent failures and suggests fixes.Agnost publishes a free plan and a plain price list for turning an agent's live conversations into fixes and tests, so how much "learning from use" you get at each price can be tried by hand.Pre-Seed
03BelvedirPost-trainingTrains a small open model on an agent’s own production traces, then serves it.Belvedir's docs say how many traces a training run needs and how a result is scored against the base model, so its claim that a small tuned model can replace a large one can be tested on traces we generate ourselves.3 open roles
04CajalVerificationMachine-checked proofs, in Lean, that compiled code does what its spec says.Cajal publishes the code of the engine its proofs run on and a worked example that builds from the repository, so the claim that a program has been proven correct can be checked by hand.
05cubicVerificationAn AI reviewer that checks each GitHub pull request against the whole codebase.cubic publishes a benchmark score and a free plan for pull-request review, so its claim to catch real bugs in AI-written code can be checked by hand on a repository of our own.1 open role
06Discovered MaterialsAI-run labsAI agents that propose new chip materials, plus a public benchmark of models doing it.Discovered Materials publishes its benchmark scores and an interactive thermal model, so its claims about AI finding chip materials can be read and checked against the numbers it shows.Seed3 open roles
07Distil LabsPost-trainingTurns your examples or traffic into small task-specific models you own.Distil Labs gives two free training runs and a command-line tool, and publishes its benchmark and example repos, so its claim that a small tuned model can match a big one can be tested on a task of our own.Pre-Seed1 open role
08Guide LabsInterpretabilityAn AI model built to show which concepts and training data led to each answer.Guide Labs publishes an 8-billion-parameter model built to show which concepts and training data produced each answer, with code and weights anyone with a large enough graphics card can download and test.Seed

The research index / 793 dossiers

Every company we cover

Search a name or ticker, then sort and triage dossiers by coverage freshness, our conviction and trading relevance. The roster above is a subset of this list.

Loading companies…