LLM / AI on top of MongoDB with Atlas

[Rasmus_Gregersen (Rasmus G.)]

Hi guys,

Has anyone tried to build an LLM, ChatGPT or similar, on top of MongoDB?
So its possible to “chat” with the database.

[Prakul_Agarwal]

Hello Rasmus,

Here how you can do that

• using a framework called LlamaIndex - Build a ChatGPT with your Private Data using LlamaIndex and MongoDB | by Jerry Liu | LlamaIndex Blog | Medium
• using python and Atlas vector search directly - Building Generative AI Applications Using MongoDB: Harnessing the Power of Atlas Vector Search and Open Source Models | MongoDB
• Something similar with framework called langchain - GitHub - mongodb-developer/atlas-vector-search-langchain: This Repo shows how to integrate LangChain, Open AI and store embeddings in the MongoDB Atlas and run a similarity search using MongoDB Atlas Vector Search.

I would be interested in learning more about what you are exploring

[Rasmus_Gregersen (Rasmus G.)]

Hi Prakul,

Awesome thanks a lot, good read

Im my case I have been working with LangChain+MongoDB Atlas+GraphQL to use ChatGPT to write a GQL query and then execute to get an output.

The goal is to be able to ask “What is the avg. sales in August?” for structured data in MongoDB.

I have been testing with MDB with GQL, but my challenge is that I haven’t found a way/solution to add the auth for the API in LangChain. I have followed the guide in LangChain: GraphQL | 🦜️🔗 Langchain

I can send my script in a message for inspo and interest?

[Rasmus_Gregersen (Rasmus G.)]

Solution for the auth in the header:

Help needed to add GraphQL API keys in LangChain script

Found a solution to connect to MongoDB with GraphQL API auth. Download the LangChain package: GitHub - langchain-ai/langchain: Building applications with LLMs through composability Edit the load_tools.py file so it can take a “headers” input (apiKey: “xzy”) pip install the package in your Python script/terminal [load_tools]