abbey  by goodreasonai

Self-hosted AI interface with document chats, YouTube chats, and more

created 9 months ago
396 stars

Top 74.0% on sourcepulse

GitHubView on GitHub
Project Summary

Abbey is a self-hosted, configurable AI interface designed for individuals and teams to interact with various AI models. It offers features like document and YouTube chats, workspaces, and supports multiple LLM, TTS, and OCR providers, enabling private, customized AI workflows.

How It Works

Abbey utilizes a Docker-based architecture, orchestrating backend services, a frontend, and a database. Users configure their preferred AI models (LLMs, embedding, TTS, OCR) and integrations (search engines, authentication, storage) via .env and settings.yml files. This declarative configuration allows Abbey to dynamically integrate with a wide range of AI services, including OpenAI, Anthropic, Ollama, and various OpenAI-compatible APIs, providing flexibility and extensibility.

Quick Start & Requirements

  • Install: Clone the repository and run docker compose up (development) or docker compose -f docker-compose.prod.yml up --build (production).
  • Prerequisites: Docker and Docker Compose.
  • Configuration: Requires .env for secrets (API keys, database password) and settings.yml for model and integration configuration.
  • Docs: Setup and Install

Highlighted Details

  • Supports multiple LLM providers including OpenAI, Anthropic, Ollama, and OpenAI-compatible APIs.
  • Integrates with various embedding models, TTS services (OpenAI, ElevenLabs), OCR (Mathpix), and search engines (Bing, SearXNG).
  • Offers flexible authentication options: Google, GitHub, Keycloak.
  • Includes an experimental web crawling feature using Playwright.

Maintenance & Community

  • Actively maintained with direct contact encouraged for issues and contributions.
  • Community engagement via Twitter DM and email.

Licensing & Compatibility

  • The repository does not explicitly state a license in the provided README.

Limitations & Caveats

  • The experimental web crawling feature has no security guarantees.
  • Users must ensure their SearXNG instance is configured to allow JSON output for integration.
  • Customization of ports requires modifying Docker Compose files.
Health Check
Last commit

3 months ago

Responsiveness

1 day

Pull Requests (30d)
0
Issues (30d)
0
Star History
34 stars in the last 90 days

Explore Similar Projects

Starred by Andrej Karpathy Andrej Karpathy(Founder of Eureka Labs; Formerly at Tesla, OpenAI; Author of CS 231n), Alex Cheema Alex Cheema(Cofounder of EXO Labs), and
3 more.

Perplexica by ItzCrazyKns

0.3%
23k
AI-powered search engine alternative
created 1 year ago
updated 1 day ago
Feedback? Help us improve.