Ai assistant with large data set

Ask for Help
1

Hey all, wondering if anyone has ever tinkered or acheived somethign similar to the below

I have a car quoting platform built in glide and the cars data file is 3500 rows

with 40+ columns (not all relevent for this usecase for the assistant) but some of the columns are decently lengthed arrays of standard equipment or additional extras

i want to create an ai agent that can help our sales team find the right car for their customer

I know this can be done logically with filters and buttons etc and maybe i will go that way but i want it to look and feel like an ai bot

I was thinking it can ask some questions first of all to filter down the 3.5k rows to a couple hundred and then maybe using the open ai ask a questions bout a table but it just seems a little clunky?

Anyone thoughts?

or even somethign outside of glide that can be embedded Or can call its api to function that way?

2

I would suggest playing around with Pinecone.

I don’t have much experience with indexes in Pinecone, but you can probably sync your Glide CRUD with a Pinecone index, and then for questions in the app, do a Query Data action to get the right row(s), feed it as context for Generate Text, and get an answer.

Another way is creating (though massive) a JSON file of all rows in your Cars table, upload it as a file to an Assistant in Pinecone, and then use Send Message to Pinecone action to interact with that assistant.