We are building private, local AI, and we want you to try it

One thing that keeps coming up in my conversations with professionals, teams and organisations about adopting AI is concern about privacy, security and protecting intellectual property.

It becomes very real when you think about the kinds of information people actually want AI to help with:

• Financial data
• Medical or health information
• Proprietary research
• Internal company documents
• Client information
• Strategy documents
• Unreleased product ideas

These are exactly the contexts where sending data to an external AI system may introduce privacy, security, compliance or intellectual property concerns.

Run more of your AI locally

One way to reduce that risk is to run more of your AI on your own machine.

Running generative AI locally is already possible. The problem is that getting a useful local setup running can still take a lot of work: finding and configuring models, connecting different tools, setting up retrieval over your files, and getting everything to work together.

Over the past few months we have been building a fully local and private version of Agent Commons, with generative AI running directly on your computer.

What is in the preview

The Agent Commons desktop app is now publicly available as an early preview. Out of the box it brings together:

• Local models for text, audio transcription, voice and image generation
• Built-in retrieval over your files
• A workspace library for your files and AI-generated outputs
• The option to install and switch between other models with minimal setup

When we say local, you can test it yourself: switch off your Wi-Fi and keep using it. Whenever you need the cloud, you can continue your work there too.

What comes next

It is still early, and there is plenty we are experimenting with and improving. The aim is to make private AI easier to set up and use, while giving people more control over their data, models, agents and workflows.

We are also exploring LAN AI, where teams and organisations use and collaborate around AI systems within their own local network, keeping data inside that environment. If that would be useful for your team, tell us.

Download the Agent Commons desktop preview, put the local setup to the test, and let us know what you think.