Don’t know if the third part relates to openAI (which is where i wanted to send my files), but the solution i’m using right now is via

  1. create upload (establish the filename)
  2. add upload part (via the s3 pre-signed URL)
  3. finish the upload (capture the file_id)
  4. connect to relevant vector store…

so got the filename issue out of the way…

Bubble should however create a feature where the existing S3 can be directly accessed via the pre-signed URL function that would significantly reduce the hassle!

1 Like