Restai

RESTai is an open source, self hosted AI as a Service platform released under the Apache 2.0 license. It puts retrieval augmented generation, agents, Model Context Protocol tools, visual logic and an app builder behind a single REST API, together with a full web interface, so an organization can run its own AI service on its own infrastructure. The platform supports five project types, including RAG, agents, a block based visual logic editor and direct inference. RAG projects work with vector stores such as ChromaDB, PGVector, Weaviate and Pinecone, offer reranking and natural language to SQL, and can sync knowledge from S3, Confluence, SharePoint and Google Drive. Agents can connect to any MCP server over HTTP, SSE or stdio. It works with more than eleven model providers, including OpenAI, Anthropic, Ollama, Gemini, Groq, Azure and AWS Bedrock, plus any OpenAI compatible endpoint. Operational features include a dashboard for token usage, cost and latency, teams with role based access control, OAuth, LDAP and OIDC sign in, two factor authentication, guardrails, audit logging, per project rate limits and budget caps, an evaluation framework, prompt versioning, white label branding and a WordPress integration. A live demo is available on the site. The software is free, and the vendor does not publish any paid plans or prices.