Judgement as a moat
A conversation with Ilse Kroon, who started Thatcham, and Ellerbeck Capital, who led the $9.4M seed.
Explore<0.42 S>
There can still be a world of craft, restraint and taste.
That means cracking how to measure, sort, search across and codify subjective domains into data a model can learn from, and tools an agent can hold.
So that models and agents return work that is not merely correct, but work a person would actually choose.
We are making the unverifiable verifiable, starting with design.
Thatcham across the stack
We work with frontier labs to give their models judgement, restraint and a working sense of craft.
We work with app teams, coding agents and creative tooling companies, supplying the context, eval and verification layer so what ships lands on brief and on brand.
Bring your eye to post-training, retrieval, evals, data, product and design, and help pin down the parts nobody has measured yet.
We are half research lab, half infrastructure company. This is what is on our mind.
Explore writing
A conversation with Ilse Kroon, who started Thatcham, and Ellerbeck Capital, who led the $9.4M seed.
Explore
Subjective judgement is the next frontier. Here is why Ellerbeck backed us to go after it first.
ExploreMaking things got cheap; deciding what is good did not. Taste, not generation, is the open problem.
Explore→ Drag a tile
Keep it04Not for me