Recommendations/Matching Function Based on User Reponse


I’m working on a feature in my app that will provide accurate recommendations to users based on their responses on a questionnaire/survey. The concept is it will take the user’s responses and provide him or her with several other users in the database that have similar responses, sort of like a matching/dating game.

I’m looking for a way to do this easily such as installing a plug-in or using external AI source, but haven’t found one that’s free/cheap (trying to keep this on a low budgt). I came across ‘RAPI - Recommendations API’ that was a plug-in introduced to Bubble in 2020, but discontinued in 2021 due to high server cost. Is there anything else out there (or another effective/efficient way internal to Bubble) that anyone can recommend I use??


Generate a list of keywords for each user and save it to the user type then Search for users based on keywords. Not perfect because one user might have keyword ‘football’ and another ‘soccer’ and they might not match when they should (because football and soccer are the same thing in civilised countries)

You can also use Pinecone’s free plan if you want. Use OpenAI Embeddings API to embed a user’s profile (e.g description, interests, hobbies) and stick it in a Pinecone database. Save the user unique ID as metadata on each vector. Then to match a user, just embed the user’s profile, and search for similar users. You can use the unique ID on each vector to find the relevant user in your database.

Thanks so much, I will give these a try!