We build and maintain open source Python tools used by AI engineers worldwide. These tools solve the hardest problems in production RAG pipelines - from document parsing and chatbot frameworks to hallucination detection. Free, MIT licensed.
LongParser handles document ingestion. LongTrainer builds the chatbot. LongTracer verifies the output. Together they cover the full RAG pipeline.
Production-Ready RAG Framework
Build multi-tenant RAG chatbots with persistent memory, streaming, tool calling, and 9 vector DB providers. 10 lines of code to a production chatbot.
pip install longtrainer- MIT LicenseRAG Hallucination Detection
Detect hallucinations in LLM responses by verifying every claim against your source documents. STS + NLI pipeline. No LLM dependency. Just strings in, trust score out.
pip install longtracer- MIT LicenseDocument Intelligence for RAG
Parse PDFs, DOCX, PPTX, XLSX, and CSV into validated, AI-ready chunks. 6 hybrid chunking strategies, HITL review, citation validation, LaTeX support, and RTL languages.
pip install "longparser[gpu]"- MIT LicenseEvery tool in this section was built because we needed it for a client project and couldn't find a good open source solution. LongTrainer came from building multi-tenant RAG chatbots. LongParser from dealing with complex enterprise documents. LongTracer from needing to verify AI outputs in regulated industries.
We open source them because the AI community benefits from shared infrastructure, and because the best way to prove technical depth is to show your work.
Use in commercial projects, modify freely, no restrictions.
Built for real workloads, not tutorials or demos.
Regular releases, bug fixes, and new features.
Integrates with LangChain, LlamaIndex, and standard Python.
We use these tools in production for our clients. Schedule a free consultation to discuss your RAG requirements.