The AI workspace that runs on your machine
Isonomic Chat is a private AI workspace for the Mac: local models by default, artifacts you can use, and projects that keep your files and context together.

Isonomic Chat
Your conversations stay yours
Chat runs on Isonomic Desktop. Conversations, files, and memory stay on your machine unless you turn on a remote model.
Local models first. Any model you want.
Run open models on your own hardware, or add your API keys for hosted ones. Compare answers and switch without losing context.

Create anything with Artifacts.
Turn conversations into interactive charts, dashboards, applications, visualizations, and working prototypes, with editable code and live previews.

Research deeply. Understand everything.
Investigate complex questions using web research, uploaded documents, connected sources, and multi-step analysis, with verifiable citations.

Your entire world, in context.
Persistent projects combine conversations, knowledge, files, instructions, and connected applications into one context-aware AI workspace.

Agents
Build assistants around your tasks, tools, and knowledge. Give each agent a clear role and let it handle research, writing, or everyday work.
Code Interpreter
Run code directly in your conversation to explore data and solve problems. Turn uploaded files into calculations, charts, and useful answers.
Artifacts
Create documents, code, and interactive previews alongside your chat. Refine the result in a dedicated workspace without losing the conversation.
MCP
Connect your AI to the tools and data you already use through MCP. Bring relevant context into your chat and take action across connected services.
Memory
Keep useful preferences and context across conversations. Spend less time repeating yourself as your assistant learns how you like to work.
Web Search
Search the web from your conversation for timely answers. Explore sources and follow citations to check the facts and dig deeper into any topic.
Transparent, verifiable, and open source
Privacy you can check, not a promise. Read the code, rebuild it from source, and build on it.


