OwnPilot.
GitHub
Your first flightInstall OwnPilot, connect a model, and give your assistant its first task.InstallationDocker, source development, prerequisites and first-run checks.A system you can understandFollow a request from your interface through the gateway to models, tools and storage.Meet your working crewChat agents, Claw runtimes, Soul identities and multi-agent crews.Give your AI useful handsBuilt-in tools, MCP servers, extensions, skills and your own CLI tools.Your choice of intelligenceCloud providers, aggregators, local models and routing strategies.From a request to a routineVisual workflows, scheduled tasks, triggers, plans and proactive autonomy.Talk where you already areWeb chat, Telegram, WhatsApp, voice and real-time feedback.A memory for your working lifeNotes, tasks, goals, contacts, custom tables and persistent memory.Ownership, with clear boundariesData flow, authentication, tool approvals, isolation and deployment responsibilities.Make it your ownEnvironment variables, Config Center, model settings and safe defaults.Give OwnPilot a reliable homeVolumes, HTTPS, authentication, backups and upgrading a self-hosted instance.Build around the gatewayHTTP APIs, streamed responses, WebSocket events and MCP.
OWNPILOT DOCUMENTATIONESC to close
v0.8.3A little more autonomous.
THE PERSONAL AI, ACTUALLY PERSONAL.

Your AI.
Your world.
Your rules.

An assistant that connects your models, tools, and memory. Runs on your infrastructure. And works on your terms.

Make it yoursExplore the source
Self-hostedMIT licensedCloud or local AI
YOUR OWN INTELLIGENCE LAYER01 — 03
One connected system. You’re in the pilot’s seat.ILLUSTRATED FLOW
YOUR INTELLIGENCE.
YOUR CHOICE.
◎ OpenAIAnthropic✦ Gemini◉ ollama↗ OpenRouter
And your favorite.
01 / BUILT AROUND YOUR EVERYDAY

Less prompting.
More getting things done.

You don’t need another chat window.
You need an assistant with context,
capability, and a little initiative.

01 /

A crew that gets you.

Give agents a purpose, a personality, and a place to work. Run a focused task or build a crew that collaborates.

Meet your agents
02 /

Context that sticks.

Notes, goals, preferences, and persistent memory. Your assistant can pick up the thread instead of starting over.

Explore personal memory
03 /

Make “every day” effortless.

Connect tools in visual workflows. Set schedules, react to events, and turn recurring work into a routine.

Build a workflow
04 /

The tools are already here.

Search the web, work with files, run code, or connect an MCP server. Extend your assistant with skills and your own tools.

Open the toolbox
05 /

Good company. Anywhere.

Start in the web UI. Keep the conversation going on Telegram or WhatsApp. Add voice when typing can wait.

Connect your channels
06 /

You hold the controls.

Your deployment, your model, your permissions. Set approval requirements and boundaries for what your agents can do.

Understand the boundaries
02 / A HOME FOR THE WHOLE PICTURE

Meet your new
command center.

Conversations, agents, memory, and workflows.
Connected in one thoughtful workspace.

Take a closer look
localhost:8080INTERFACE WALKTHROUGH
Workspace / Autonomous hub
YOUR SPACE TO GET THINGS DONE

A small crew. A clear mission.

Give every agent a purpose, a workspace, and boundaries.

Your agents
Your infrastructureYour permissionsYour choice of model
Scheduled

Morning briefer

A calmer start to your day.

On your schedule
Ready

Research assistant

Follow the sources. Find the signal.

Within your boundaries
Ready

Code reviewer

A fresh pair of eyes on your work.

Within your boundaries
A workflow, in plain sight.ILLUSTRATIVE EXAMPLE
  1. Review the day’s contextCalendar & tasks
  2. Prepare a focused briefingYour chosen model
  3. Deliver it where you workTelegram
ONE PLACE. ALL THE MOVING PARTS.Illustrated from the product’s documented features. Sample content.
03 / CONNECTED BY DESIGN

Many moving parts.
One clear direction.

OwnPilot brings your AI stack together. You choose what connects, what runs, and how much freedom it gets.

  1. 01

    Choose your intelligence.

    Connect a cloud provider or a local model. Switch without rebuilding your workflow.

  2. 02

    Give it the right context.

    Connect tools, memories, and instructions. Your assistant has something useful to work with.

  3. 03

    Set the boundaries. Let it work.

    Start a conversation, schedule an agent, or run a workflow. Stay in control of the next step.

Explore the architecture
Web UI➤ Telegram◉ WhatsApp⌁ MCP clients
OwnPilot gatewayHono · HTTP / SSE / WebSocket
:8080
Agent engineProvider routerTool orchestrationEventBus
Your modelsCloud APIs · local inference
Your toolsBuilt-in · MCP · extensions
Your dataPostgreSQL · memory

Run the gateway and database on your infrastructure. External services follow your configuration.

OWN YOUR ASSISTANT. REALLY.

Intelligence shouldn’t
come with a landlord.

Run OwnPilot where you want. Keep your application data on your infrastructure. Choose the models and services you trust.

A clear look at privacy & security
Your infrastructure.

Self-host the gateway and PostgreSQL database.

Your boundaries.

Configure tool approvals, autonomy, and budgets.

Your freedom to build.

Inspect, extend, and contribute under the MIT license.

Cloud providers and connected tools may receive data when used. Local inference and self-hosting give you choices; review each connection.
04 / START WITH SOMETHING USEFUL

A little help.
A lot of room to explore.

Give it a job that matters to you.
Build from there.

01 / THE MORNING PERSON

Your day, already in focus.

Bring together your calendar, tasks, and a short briefing. Deliver it to the channel you actually check.

SCHEDULES·PERSONAL DATA·CHANNELS02 / THE CURIOUS MIND

Turn curiosity into clarity.

Ask a research agent to explore a topic, gather sources, and save a summary as a versioned artifact.

BROWSER·AGENTS·ARTIFACTS03 / THE BUILDER

Your next coding companion.

Coordinate coding CLIs, follow terminal output, and let an agent help investigate or review a change.

CODING AGENTS·CLI·WORKSPACES

Example workflows. Configure the required models, tools, accounts, and permissions in your instance.

05 / YOUR NEXT GOOD DECISION

A few commands.
A world of your own.

Bring Git and Docker. We’ll bring the assistant.
Clone, configure, and make yourself at home.

Docker + ComposeGit

Copy the environment template and set a strong database password before starting. Configure your AI provider in Config Center after launch.

All installation options
terminal · macOS / Linux
git clone https://github.com/ownpilot/OwnPilot.git ownpilot
cd ownpilot
cp .env.example .env
# Set a strong POSTGRES_PASSWORD in .env before starting.
docker compose --profile postgres up -d --build
Next: open localhost:8080Full setup guide
A FEW THINGS, CLEARED UP.

Good questions.
Straight answers.

Still curious about something?

Let’s talk on GitHub
01

What exactly is OwnPilot?

OwnPilot is a self-hosted personal AI assistant platform. It connects your choice of AI models to tools, persistent memory, autonomous agents and workflows through a web interface, Telegram or WhatsApp.

02

Can I use it with local models?

Yes. OwnPilot supports Ollama, LM Studio, LocalAI and vLLM, alongside cloud providers and other OpenAI-compatible endpoints. Model capabilities and tool-calling support vary; choose a model suited to your tasks.

03

Does self-hosted mean my data never leaves my machine?

Your application and database run on infrastructure you control. Requests to cloud AI providers, connected apps and external tools can send data to those services. Use a local model and review your enabled integrations when you need a local inference workflow.

04

What do I need to get started?

For the Docker path, install Git and Docker with Compose. Clone the repository, copy .env.example to .env, set a strong database password and start the postgres profile. Open localhost:8080 and configure a provider in Settings → Config Center. Source development requires Node.js 22+, pnpm 10+ and PostgreSQL 16+.

05

Can agents act without my approval?

OwnPilot offers configurable autonomy levels, tool policies, approval requirements and budgets. Claw agents can run a single task, continuously, on an interval or in response to events. Review the permissions and limits for each agent before enabling autonomous execution.

06

Is OwnPilot free to use?

OwnPilot is open source under the MIT license, including for commercial use. Your hosting, paid model APIs and third-party services may have their own costs. OwnPilot does not include free access to those services.

07

Can I connect my own tools?

Yes. Connect external MCP servers, install skills in the SKILL.md format, add extension bundles or register CLI tools. OwnPilot also exposes its own tools through an MCP server endpoint. Review permissions before installing third-party tools.

08

How can I contribute?

Read CONTRIBUTING.md in the product repository, check the open issues and propose a focused change. Bug reports, documentation improvements, extensions and pull requests are welcome. Website contributions belong in ownpilot/ownpilot-site.

LESS DEPENDENCE. MORE POSSIBILITY.

Put yourself
in the pilot’s seat.

Your models. Your tools. Your next great idea.

Let’s make it yoursBuild with us
OPEN SOURCE. SELF-HOSTED. ENTIRELY UP TO YOU.