Custom software development for businessPay for results
Fix incorrect answers in a LangChain, OpenAI and Pinecone support chatbot
Project brief
An existing SaaS support chatbot uses LangChain and OpenAI with roughly 300 help-center articles in Pinecone. It is returning wrong answers, or claiming information is missing even when the documentation contains it.
Find the cause, whether retrieval, chunking, embeddings or prompting, and correct it. There are 15 failing questions to test against. Include a short explanation of what went wrong. This is a repair, not a new chatbot. Authorized private-repository access, documents, API access and the expected answers will be agreed before starting. Please open your proposal with your likely diagnosis and relevant RAG experience.
Deliverables & acceptance
What you'll deliver
- Focused code/configuration correction, test results and root-cause explanation.
What the result must meet
- All 15 agreed questions are checked before and after the repair; final answers agree with the supplied documentation.