Your work
stays yours.

A private AI workspace that runs on your hardware. Bring your files, agents, and workflows together. Keep control of where your work goes.

Your hardware. Your models. Your data.
localhost:3000

Client research

Project overview
Summarize the key findings from the selected documents.

The selected documents indicate three consistent findings:

  1. Clients need clear answers they can trace to original research.
  2. Repeated tasks benefit from an agreed process and a review step.
  3. Confidential files need a defined boundary for access and use.
Ask about this project…
Local model · Running on your machine

Example project workspace

Private by default
Grounded in your files
Built for repeatable work

A little structure.
A lot more clarity.

Give your AI the context, boundaries, and process your work deserves.

Everything your project needs. Nothing it doesn’t.

Keep files, instructions, and recorded decisions in one place. Source links help you check the material behind an answer.

Explore workspaces
Client research
Research notes.pdf124 KB
Interview summary.docx38 KB
Project instructions.md2 KB

The right role.
The right permissions.

Give each agent a specific job. Define what it can read, what it can write, and which other agents it can request.

Meet your agents
Read project filesSelected files
Write project filesNot allowed
Request another agentReader only
Web accessOff

Good work, made repeatable.

Save the steps from input to output. Put human review where it matters, then use the same process next time.

See workflows
Read
documents
Human
review
Write
brief

Your model. Your machine.

Choose an open-weight model that fits your hardware. Set concurrent runs and decide when to keep a model loaded.

Find your setup
Local model
Model size8B parameters
Concurrent runs2
Keep loadedOn

From your files
to your next step.

01

Choose your home

Install on your computer or a private server. Choose a local model that fits your hardware.

02

Give your project context

Add the files, instructions, and decisions the work needs. Set permissions for each agent.

03

Run, review, repeat

Start a workflow, review the important steps, and save a result you can trace to its sources.

A clear line
around your data.

Local inference keeps project content on your hardware. Anything that reaches outside is a separate decision.

Understand the data boundary
Your computer or private server
Project filesLocal modelYour output

Outside your boundary

Web researchOff by default
External connectorsOff by default

Make room for work that stays yours.

Explore a private workspace built around your files and your decisions.

Source material

Sample source material for this product walkthrough.