AI applications that keep you in charge and your data private.
We build AI assistants for individuals and teams. The assistant proposes, you decide, and your data stays where it already lives.
Two promises, in everything we build.
You stay in charge.
Our AI suggests and explains; a person approves before anything happens. Nothing goes out in your name unchecked.
Your data stays private.
We favor models you can run on hardware you control, and we work within the permissions your systems already enforce instead of copying data somewhere new.
What we've built
Open-source projects and free tools that put both promises into practice. They're how we prove the approach.
ThetaTerm
Plain English in, the right shell command out. Runs on a small local model and asks before anything runs.
Open source, Apache-2.0.
ThetaServe
One command to launch an open-weight model on a rented GPU and get back a secured, OpenAI-compatible endpoint. Pick a model and a use case; thetaserve finds the cheapest GPU that fits.
Open source, Apache-2.0.
LLM Sizing Calculator
Check whether a local LLM fits your machine and how fast it will run. Look up any Hugging Face model, right in your browser.
Closed source.
Custom assistants for your business.
We build assistants around your systems, your workflows, and your rules, from a proof of concept with real users through to onboarding and support. What we learn from each build makes the next one better.
Have a workflow in mind? Let's talk it through.
A 30-minute conversation about the work you want help with, the systems it touches, and what has to stay private. No pitch deck, no pressure.