Cockpit

Windows · macOS · Mobile clients

Your AI. Your projects. Your command center.

Move from conversation to practical work. Cockpit brings your models, projects and machines together in an interface that keeps context close at hand.

In development · Ask about access and availability

Cockpit
Demo project

Conversation

Read-only
You

Summarize the project documents and suggest next steps.

Cockpit

I will prepare a summary from the attached documents. File changes will require a separate request.

Project notes.pdfPDF · Attached text
✳ Project context

Instructions, documents and conversations in one place.

Illustrative preview · Fictional project, machines and conversations
01Your choice of AI02Context for each project03Your connected machines

01 / One workspace

A thread for your ideas.
Tools to move them forward.

Chat is the starting point. Projects, documents and tools give it a working context.

01

Conversations, organized

Group conversations by project and find their associated files. Return to a topic with its documents and context.

02

Models, your way

Choose from your AI connections: Anthropic, OpenAI, xAI, DeepSeek or a compatible API. CLI engines and local models have their own workflows.

03

Documents in your chat

Attach a text-based PDF, Word document or text file to ask for a summary or comparison. Attachments stay associated with the conversation.

04

Memories that fit your work

Activate instructions, adjust response style and organize useful knowledge. Personal settings and project settings have distinct scopes.

05

Machines and their tools

Find your devices, files and terminal sessions. Machine control requires a configured agent, authorized folders and a compatible engine.

06

Work you can follow

Review executions, permission requests and scheduled wake-ups. Keep track of what starts and what is waiting for your approval.

02 / Your working hub

One Cockpit.
Several devices.

A Mac or PC can host your personal Hub: it stores projects and coordinates connections. Your other devices join through explicit pairing.

The Hub must stay powered on and reachable. Connecting across networks requires a suitable connection. Cockpit does not automatically open router ports.

Your computer
Personal HubCockpit
Paired clients

03 / Clear capabilities

Chat and action,
with the right engine.

Chat with your providers

“Mes IA” connections use your API keys and send text to your chosen provider. This chat workflow does not run tools on your machines.

Work on your machines

Compatible engines can use available tools within the authorized scope. The local agent, session rights and permissions still apply.

Provider subscriptions, keys, models and costs remain your responsibility. A chat-service subscription does not automatically include API credit.

04 / In practice

A few things to know

Can I download Cockpit here?

Not at the moment. Cockpit is in development and this page introduces it. Email me about trial possibilities; no purchase or public download is offered here.

Which devices does it run on?

The project has Windows, macOS and iPhone clients. Capabilities depend on the client and Hub versions; Linux packages and compatibility still require qualification.

Does my data stay on my computer?

Projects and conversations are stored on your selected Hub. Text sent to AI, including text from attachments, is transmitted to your selected engine or provider. Personal hosting does not mean all inference runs locally.

Can every model control machines?

No. Models configured in Mes IA provide text chat. Machine control uses a separate workflow with a compatible engine, an agent and the required permissions.

Can teams already use it together?

Member sharing is part of the commercial roadmap, but its availability should not be assumed. Personal pairing described here connects the owner’s own devices.

A workspace for your projects.
A place for your AI.

Want to discover Cockpit or discuss a specific use case? Get in touch.

Let’s talk about Cockpit ↗← All products