A lawyer building privacy-first legal AI

I trained as a lawyer. What pulled me toward technology was a plain mismatch: the firms I saw wanted to put the best AI to work on drafting, review, and research, and could not, because doing it meant handing privileged client documents to a third party. The value was obvious. The risk was unacceptable. Most tools asked firms to accept the second to get the first.

Confidentiality should be a property of the architecture, not a promise in a policy.

So I built inchambers.ai the other way around. Documents are processed on the user's own device and encrypted there. A firm brings its own AI keys, or runs the whole platform on its own infrastructure. The files, the keys, and the prompts never reach our servers, because there is nowhere on our side for them to land. A firm gets frontier models for the work that matters, and keeps the one thing it cannot afford to give up.

The platform lives where a lawyer already works: inside the browser and inside Microsoft Word. The firm chooses the models and decides where its data lives. My job is to make the private way the easy way.