I am looking for a Retrieval-Augmented Generation (RAG) tool that can integrate with Make.com via an API. My goal is to build a chatbot that uses RAG to provide accurate and context-related answers based on a knowledge database.
Specifically, I want to achieve the following functionality:
The user’s answers are forwarded as JSON to Make.com via a webhook.
In Make.com, the answers are processed and stored in a database for further processing.
I cannot find a tool that already has an API integration for this. Is this correct? And is there perhaps someone who has already built this themselves?
Hey @Robert_Tellier and @lawops - As someone who has built out a number of RAG apps using all sorts of different vector databases as backing (Pinecone, Elastic/OpenSearch, PgVector on Postgres), I can pretty confidently say that I don’t think make is the right tool for the job… unless you’re integrating with an off the self solution that offers this functionality out of the box.
There are a lot of products that let you build RAG apps with a pretty simple interface (that you pay a premium for) but that enable you to get off the ground and running very easily. There are also some no-code / low-code solutions that are dedicated to making building out RAG apps much easier such as Flowise that may be a good path to go down.
Otherwise, to build a RAG apps, you’ll likely find yourself in a more code focused implementation with tools like LangChain which is great, but again much more tech heavy / developer oriented. Most of my experience, admittedly, is from a developer prospective using LangChain, but given how new the technology is, it’s going to take some time for tools like make and others to directly catch up.
If you have more questions, I’m happy to try and help and point in the best direction. Feel free to comment here or reach out directly!
We’re building exactly this at Dumpling AI (dumplingai.com)
We have a RAG tool that is usable as a module in Make.com.
Basically, you can upload your data (PDFs, documents, etc.) to our platform, then in the Make.com module you can pass in the user query and get relevant chunks back for you to use in Make.com (e.g. to then pass into your LLM or to process first).
We’re in private beta at the moment, but feel free to sign up for early access and submit your use case.