OpenMake / 1.13.0

Your AI. Your server. Your rules.

A local-first AI platform for teams that want a usable chat workspace, autonomous work, research, tools, and model choice without giving up operational control.

Choose a path

Evaluate now or run it yourself

Use the hosted demo or the macOS Desktop app when you want to see the workspace quickly. Self-host when your team needs its own gateway, storage, model policy, and network boundary.

  • Try now: Live Demo or macOS Desktop
  • Run your own: Node 24, Docker, PostgreSQL, and an OpenAI-compatible endpoint

Real product

One workspace for the work around a model

OpenMake keeps chat, agent tasks, deep research, MCP tools, artifacts, model roles, and data controls in one place. The default chat route is local-first; external connections are explicitly configured.

  • 262K effective context-fit window
  • 22 built-in MCP tools
  • 100 routed specialists across 18 domains
  • Korean, English, Japanese, and Simplified Chinese interface languages

Models and connections

Keep a local route. Add external capacity deliberately.

Start with your own vLLM or LiteLLM-compatible endpoint. Add a provider only where it earns its place, and keep credentials under your operating policy.

  • OpenRouter API key
  • ChatGPT Plus or Pro OAuth device flow, unofficial and policy-dependent
  • Ollama Local and Ollama Cloud
  • NVIDIA NIM

Trust and control

Designed for an operator, not a black box

Docker keeps service roles separated. Role and audit controls make model access reviewable. The Desktop bridge scopes local work to a connected folder and asks before remaining shell commands.

  • Pending database migrations run at server startup by default
  • Set DB_AUTO_MIGRATE=false when your operation requires manual migration
  • Desktop updater verifies SHA-256 before replacement

Start

Pick the next useful step

The demo is the fastest way to judge the product. The installation guide is the shortest route to an operator-owned environment.