Local AI setup
A local AI setup puts AI on your own machines, for founders and teams who want their files and their data to stay with them, with the open models that suit their work and the hardware they have.
Where you are
You want AI working for you without your files, your clients' data or your ideas leaving your own machines. Then you open the list of open models and find more every week, some with guardrails and some without, each in sizes and compressed versions that run differently on a laptop, a desktop with a graphics card or a server, and nothing on the list says which one fits your work.
What you get
- Open models chosen for the work you do, whether that's writing, code, images, voice or answering questions from your own documents.
- Each one in the size and compressed version that runs well on the hardware you have, or on the machine you plan to buy, from a laptop to a server your whole team shares.
- Guardrails set for the work, careful where a team needs them and open where the work does.
- Hosted models such as Claude or ChatGPT alongside them, for any part of the work you'd rather give one, with your files and your operations still on your own machines.
- The whole setup written down, so you can swap a model out when a better one comes along.
How it goes
- You tell me what you want AI to do, and what machines you have or are thinking of buying.
- I pick the models and the setup that fit, and walk you through why each one is there.
- I set it up on your machines and tune it on your real work.
- The setup is written down, and you run it from there.
The work
The first reply names the models and the setup that fit your work, on the hardware you have or plan to buy.