DocsGPT

DocsGPT is an open-source AI platform designed for creating intelligent agents and assistants with a focus on privacy and enterprise search. It allows users to build agents using an Agent Builder and perform deep research and document analysis across a wide variety of formats, including PDF, Office documents, web content, and audio files (MP3, WAV, etc.). Key capabilities include: - Wide Format Support: Ingests data from PDF, DOCX, CSV, XLSX, EPUB, MD, RST, HTML, JSON, PPTX, and audio files. - Web & Data Integration: Supports ingestion from URLs, sitemaps, Reddit, GitHub, and web crawlers. - Speech Workflows: Includes voice input recording and audio transcription for searchable knowledge. - Flexible Model Deployment: Compatible with major LLM providers (OpenAI, Google, Anthropic) as well as local inference engines like Ollama and llama_cpp. - Integration & Tooling: Provides API connectivity for actionable tools, along with pre-built HTML/React chat widgets and bots for Discord and Telegram. - Enterprise Readiness: Supports Kubernetes deployment, OIDC/SSO login, role-based access control (RBAC), and source citations to reduce hallucinations.