RESTai
Self-hosted AI platform with a web dashboard and REST API. Connect local LLMs through Ollama or vLLM, or use cloud providers. Apache 2.0 licensed.
RESTai is a self-hosted AI platform for developers and teams building document assistants, AI agents, and services other apps can call. Each project has its own model, prompts, and tools, with a browser dashboard for testing chats and tracking usage. It's open source under Apache 2.0 and runs on your own server, including through Docker or Kubernetes.
Local LLM backends include Ollama and vLLM. Cloud connections include OpenAI, Anthropic, Gemini, and Azure, so hosting the platform yourself doesn't mean every model request stays local. Cloud-backed projects send requests to their chosen provider. RESTai also detects NVIDIA GPUs and reports their usage.
Document assistants can answer questions using uploaded files, rerank retrieved passages, and query MySQL or PostgreSQL in plain language. Knowledge bases can sync with web pages, S3, Confluence, SharePoint, and Google Drive. An optional knowledge graph connects named entities with their source documents and lets readers explore those relationships.
Agents can use MCP tools and a browser to fill forms or download files. RESTai also exposes projects to Claude Desktop and Cursor through its own MCP server. A Blockly visual editor connects projects into workflows, while OpenAI-compatible APIs cover chat, image generation, and audio transcription.
For shared deployments, teams get access controls, audit logs, and token, cost, and latency analytics. Prompt history links to evaluations so teams can compare answer quality across prompt versions. An embeddable chat widget publishes project conversations on other websites.
