Need help to search my own textfiles through a Bubble frontend using Pinecone

[anders451]

Hi!

For starters I’m completely novice in programming so I’ve been following a Youtubevideo to learn how to AI chatsearch my own textfiles using a bubble frontend. The whole concept is to help my colleagues in healthcare to easily obtain internal information and get advices on how to treat patients. What I’m trying to do now is to make an AI chatbot where you can ask for routines and specific rules locally.

I’ve created API connections to OpenAI for embedding and Pinecone for uploading to vectors. I’ve built a backend workflow and a workflow in bubble. But I still don’t get it to work, the text doesn’t reach Pinecone but I manage to make successful calls (Openvector1).

I’ve been following this video: Upload and Search Your Own Documents With Bubble, OpenAI, and Pinecone (Part 1)

I’ve reached out to the very helpful guy who made the video and he says it might be outdated.

Is there anyone who could help me correct potential mistakes I’ve made? I’m uploading pictures of the workflows below.

Thankful for any help to solve this issue. Thanks!

/Anders

[tobixzybolumole]

Hi @anders451

I have done something similar for a client i worked on a healthcare webapp, i used a plugin that was able to search through the textfiles (though mine were textfiles extracted from a doc)

Instead of using Pinecone, can’t you use OpenAI assistants to do that? as it’s able to read documents being uploaded to a bubble app

[anders451]

I’m open for any solution. Do you have a guide you could recommend?

[tobixzybolumole]

Are you familiar with using OpenAI assistants?

[anders451]

No, sorry. I’ve bought Tokens on OpenAI for using the embedding function. That’s all.

[anders451]

It’s quite difficult to catch up with this AI boom. Some say I should use the Pinecone API plugin and skip OpenAI and some say I should skip Pinecone. Is there any up-to-date solutions that is easy to follow I would be very grateful.
Thanks,

[tim12333]

@anders451

But I still don’t get it to work, the text doesn’t reach Pinecone but I manage to make successful calls

I haven’t used Pinecone, but if you’re searching for the raw text that you’ve added to the pinecone DB, it’s definitely possible that they don’t show it, and they only show the vectorized data.

If you see that “vectors” are being added, then everything should be working - you just have to query it now.

[PLUGIN] - Bubble RAG

Hi all, I’ve just released a Bubble RAG AI plugin. This one is all about using your Bubble database to store data in an appropriate format for Retrieval-Augmented Generation. It’s a relatively complex plugin, but I’ve tried to keep things as simple as possible, so it’s not too over whelming to setup. What this plugin allows you to do, is to add an AI-powered question and answer feature to your app. You upload content (plain text, PDFs, or images) and then ask questions about that content usin…

If it still doesn’t work, this plugin might be worth looking into (I haven’t tried it, but it looks solid). It uses your Bubble database to store the vectors, instead of needing to use Pinecone.

It’s also possible to use Supabase to store and query vector data - This is what I’ve done in the past.

[pork1977gm]

So that plugin will save you a few pennies for sure, you won’t have to use Pinecone at all and the actual part of it which does the similarity checking to return the relevant data, works well. Thank @tim12333

[anders451]

Nah, it sadly doesnt show up in the Pinecone DB as vector, only my API call is registered.

I will look into the plugin you suggested.

Big thanks to you all