Retrieval Augmented Generation (RAG) Support | Voters

Retrieval Augmented Generation (RAG) Support

planned

Adam Bunch

Please add support for a RAG inside projects that we can query inside them.

We use another tool for this and it's vital for our workflow to be able to query our knowledge bases without sending the entire file each time.

March 3, 2025

Mikko Parkkola

If you realize this support, please support different embedding models. Currently the best I know is Voyage AI's voyage-3-large and voyage-multimodal-3, and I would love to use them with boltai.

Mikko Parkkola

Hi, I'm using Elephas to feed the AI personal context, and I must say it makes a big difference. Context is the king. I would also love to see some simple RAG support in BoltAI, perhaps by defining one folder to watch and vectorize ("my documents" perhaps?).
At the moment I'm using VoyageAI models through their APIs for the embeddings and vectorization, and it would be fantastic to have these integrated.
Projecting to the future, perhaps even vectorizing some message content from emails and instant messages through the MCPs to have full personalization of the AI via the access to the user's most personal data and context.

lucas challamel

To me, that would be an must have to keep using BOLT AI. At the moment, for RAG, I am using MSTY on MacOS, which does a great job free of charge. And for everyday convenience AI I am using RAYCAST with Advanced AI. Come on BOLT team, we believe in you :-)

Daniel Nguyen

marked this post as

planned

Charlie Cortial

Agreed! RAG would be an absolute killer addition feature to what is BoltAI right now. I don't know if it's possible to install / integrate a database like vector database with RAG like Weaviate, but that would be a must!