bionic-gpt  by bionic-gpt

On-premise ChatGPT replacement for generative AI with data confidentiality

created 2 years ago
2,222 stars

Top 20.8% on sourcepulse

GitHubView on GitHub
Project Summary

BionicGPT offers an on-premise, self-hostable alternative to ChatGPT, prioritizing data confidentiality for enterprises and individual users. It provides a familiar chat interface, robust AI assistant creation with Retrieval Augmented Generation (RAG) capabilities, and comprehensive team management features, all while emphasizing security and observability.

How It Works

BionicGPT leverages a microservices architecture orchestrated via Kubernetes. It utilizes PostgreSQL for data storage, including vector embeddings, and supports various open-source LLMs, allowing seamless switching. The system emphasizes Retrieval Augmented Generation (RAG) for AI assistants, enabling them to process diverse document formats (PDF, HTML, PNG, etc.) through a no-code UI for configuring embeddings and chunking. Security is a core tenet, with features like Row Level Security (RLS) in Postgres, Content Security Policy (CSP), and non-root container execution.

Quick Start & Requirements

  • Installation: Docker Compose is recommended for local setup and small pilots. Kubernetes deployment is supported for scaling.
  • Prerequisites: Kubernetes cluster (k3s, Docker Desktop, or cloud), PostgreSQL, MinIO or S3-compatible storage.
  • Documentation: Homepage, Contributing, Documentation

Highlighted Details

  • Supports 80% of enterprise data formats, including PDF, HTML, CSV, PNG, and PPTX, with OCR capabilities.
  • Offers enterprise-grade security features like SAST, RLS, CSP, non-root containers, and SIEM integration.
  • AI Assistants can be exposed as OpenAI-compatible APIs with key management and throttling.
  • Integrates with Airbyte for batch data uploads from numerous sources and supports real-time data capture.

Maintenance & Community

The project is actively maintained, with a focus on regular updates and new features. Community engagement is encouraged via LinkedIn.

Licensing & Compatibility

The project appears to be licensed under the Apache 2.0 license, allowing for commercial use and integration with closed-source applications.

Limitations & Caveats

While designed for scalability, the primary deployment method highlighted is Kubernetes, which may present a higher barrier to entry for users not familiar with container orchestration. Some advanced security features like resistance to timing attacks are listed as "coming soon."

Health Check
Last commit

3 days ago

Responsiveness

1 day

Pull Requests (30d)
33
Issues (30d)
10
Star History
80 stars in the last 90 days

Explore Similar Projects

Starred by Chip Huyen Chip Huyen(Author of AI Engineering, Designing Machine Learning Systems), Daniel Han Daniel Han(Cofounder of Unsloth), and
1 more.

airweave by airweave-ai

0.6%
3k
Semantic MCP server for AI agents
created 7 months ago
updated 2 days ago
Feedback? Help us improve.