Skip to content

Releases: tesseractai-lab/TesseractRAG

TesseractRAG v2.0.1 — Production-Ready Release

Choose a tag to compare

@zeyadusf zeyadusf released this 21 May 15:50

This release focuses heavily on improving the overall user experience, conversation quality, system stability, and production readiness.

TesseractRAG is now significantly more context-aware, smoother during interactions, and more reliable under real-world usage.


✨ Highlights

🧠 Smarter Query Rewriting & Context Awareness

  • Enhanced query rewriting pipeline with conversation history support
  • Better understanding of follow-up questions and multi-turn conversations
  • Improved contextual prompt generation for more natural interactions

💬 Improved User Experience

  • More user-friendly and conversational AI behavior
  • Smoother interaction flow and response generation
  • Added automatic chat scroll-to-bottom behavior
  • Improved session handling and reduced relogin friction
  • Better refresh token experience for more stable sessions

⚡ Performance & Reliability Improvements

  • Added reranker warmup optimization for faster initial responses
  • Improved API response handling and model switching logic
  • Enhanced caching behavior without requiring Redis in all scenarios
  • Better overall responsiveness and runtime stability

👤 Guest User System Enhancements

  • Implemented improved guest account creation flow
  • Better guest session management and usability improvements

🛠 Infrastructure & Configuration Improvements

  • Added Redis URL configuration support
  • Improved limiter configuration integration
  • Removed hardcoded values in favor of environment-driven configuration
  • Fixed API base URL trailing slash issues
  • Updated production environment configuration handling

🧹 Internal Fixes & Cleanup

  • Fixed message and evaluation schema edge cases
  • Improved validation behavior across multiple components
  • General bug fixes and stability improvements
  • Updated and improved project documentation and README

🔥 Overall Result

TesseractRAG v2.0.1 delivers a noticeably more polished AI experience with:

  • Better conversational memory
  • Smoother interactions
  • Improved reliability
  • Stronger production readiness
  • More natural and human-friendly responses

The assistant now feels significantly smarter, faster, and more enjoyable to use during real conversations.

TesseractRAG v2.0.0 — Production-Ready Release

Choose a tag to compare

@zeyadusf zeyadusf released this 17 May 23:17

What's New in v2.0.0

TesseractRAG v2 is a full rewrite of the original prototype,
built for production use with a modular, extensible architecture.

Highlights

  • 🔐 JWT Authentication + Guest Access
  • 🧠 Multi-provider Embeddings & Generation (Jina, Groq, HuggingFace)
  • 🔍 Hybrid Retrieval with BM25 + Vector Search
  • 🌍 Arabic + English language support
  • 🗄️ PostgreSQL + pgvector + FAISS
  • 📄 PDF, DOCX, TXT parsing
  • 🛡️ Admin panel with role-based access
  • 📊 Evaluation pipeline (Cohere)
  • 🐳 Docker-ready with DB migrations (Alembic)

Migration from v1

See BREAKING CHANGES in the PR description.

Full Changelog: v1.0.0...v2.0.0

v1.2.0

Choose a tag to compare

@zeyadusf zeyadusf released this 16 Apr 11:23
c07efaa
Release v1.2.0 - RAG evaluation pipeline