What AI Can and Can't Do for a Law Firm Yet
Extracting facts and tracking deadlines already work well. Legal judgment and predicting outcomes don't. A practical map of seven kinds of legal work.
LAb notebook
Blogs
Updates from the lab as work happens — new results, deployment notes, and conference recaps.
Scaling Zero-Knowledge Proofs for Neural Networks
Eighteen months ago, proving a single inference pass took 4,000x longer than the inference itself. Here is how we got that down to 40x.
Why an 80% Forecast Band Covered Only 61%
On retail sales data, our forecast bands promised 80% coverage and delivered 61 to 65%. Here's why the most accurate model isn't always the most honest one.
OpenWeights DAO: A Retrospective
Our six-month pilot in community-governed model training just wrapped. Here is what worked, and what we would not do again.
Our Market Lost by 11 Points, and That's Useful
In a scripted dry run, our agent market lost to a central router by about 11 points. It says nothing yet about real AI agents, but it shows how easily market costs can swamp market gains.
Why Verifiable AI Needs Blockchain
Model cards and API terms of service are promises. We think verifiability should be a property you can check, not a promise you have to trust.