DocsGPT

DocsGPT helps teams search company documents and build AI assistants that answer from those materials. It can run inside your network with local models, including an air-gapped deployment where documents and questions stay within that network.

• Cited document answers: Answers include the source title and page number so readers can check the material behind a response.
• Connected knowledge: Search PDFs, Office files, web pages and audio, use OCR for scanned PDFs, and sync material from Google Drive, SharePoint, Confluence or S3.
• Agents and workflows: Agents can research questions and call tools through MCP or APIs, while visual workflows combine agents, conditions and code with run history and human approvals.
• Embedded assistants: Add chat or search to a website, connect messaging bots, or make an assistant's knowledge available through an OpenAI-compatible API or MCP server.
• Team controls: Single sign-on, user provisioning and role-based permissions control access; guardrails can flag, redact or block inputs and outputs, and an audit log records questions, answers and actions.

The platform is open source under MIT and supports Docker-based deployment on macOS, Windows and Linux. Local inference works with Ollama, vLLM and llama.cpp. Choosing OpenAI, Anthropic or Google sends model requests to that cloud provider, even when the platform runs on premises. Arc53 also operates cloud hosting and dedicated managed instances.