01 / Start with an idea
A thought.
A photo.
A place to start.
Ask a question, drop in an image or dictate a draft. Review it before you send, then work through the reply together.
Images need a vision-capable model.
Your open-source desktop assistant
An assistant.
On your terms.
Talk through ideas. Work with images. Connect your tools. Choose the cloud or local models that work for you.
macOS · Windows · Linux · Free & open source
Real app captures. Illustrative conversations.
A conversation that feels like yours
Type it. Show it.
Talk it through.
App appearance
A thought.
02 / Think out loud
Some ideas
sound better
out loud.
Start Live and talk it through. Interrupt, ask a follow-up and see what your tools are doing, with the transcript right there.
Hang up when you’re done. The thread stays.03 / Pick it up tomorrow
Your next thought
has a place
to come back to.
Keep separate conversations for your ideas and projects. Reopen a thread with its messages, images and saved tool results.
History is saved locally. You can turn saving off.
Make yourself at home
Your models.
Your way of working.
Choose the intelligence
Pick the model.
Keep the assistant.
Use your own provider keys, or run a local model. Choose your voice, language and instructions.
Explore supported providersBring your tools
From a question
to getting it done.
Connect services through MCP, an open standard for AI tools. Add skills and see tool activity and results in the conversation.
The weather check above is one example. The tools you connect decide what your assistant can do.
See tools and integrationsTake your voice into other apps.
A global shortcut turns speech into text wherever you’re writing.
Make room for your assistant
Meet your next
desktop companion.
The app is free and open source.
Cloud AI usage is billed by your chosen provider.
macOS
Apple Silicon · macOS 13+
Download .dmgWindows
Packaged desktop app
Download .zipLinux
Packaged desktop app
Download .tar.gzPrefer the terminal? Install with uv Recommended in the project guide
With uv installed, run:
uv tool install --python 3.12 "roomkit-ui[all]"
roomkit-uiSome optional local voice components need a separate install. Read the setup guide ↗
- 01
Open the app.
Install a package or use the uv command.
- 02
Choose your model.
Add a provider key or configure a local model in Settings.
- 03
Start a conversation.
Type a message or start Live to talk.
Before you jump in
A few things
to know.
Do I need an API key?
For a cloud provider, yes: add your own key in Settings. An all-local setup can run without a cloud API key once its models and components are installed.
Can I use it offline?
Yes, with local speech recognition, a local language model and local speech synthesis. Download the models first. Cloud providers and online tools still need an internet connection.
Where do my conversations go?
Saved history lives on your computer. Cloud providers receive the content sent to their models; connected tools receive the data needed for their actions. You choose which ones to use.
For the builders among us
Like the app?
Build what comes next.
RoomKit UI is built with the RoomKit Python framework. Explore the source, report a rough edge, or build your own conversational app.