Hi guys, very new to workflow and agent building.
I tried my best to create an AI assistant that specializes in Fitness, Finance and Productivity (for personal use).
The TLDR of this assistant is : 1 main agent that talks to other agents (subworkflows) and has access to a Pinecone index that has a lot of data about me (Journal Entries, Bank Statements, Whoop sleep performance).
And my main issue comes from that Pinecone node.
This is how my Memory workflow is set up :
This is what happens basically :
- WF 1 : Everyday the WF checks for new downloads in my Google Drive and downloads it to Pinecone (text small 3 embeddings)
- WF 2 : Also everyday, checks the messages sent by the coach on sheet, downloads it to drive and marks said message as “updated”.
I did my research on pinecone and my first thought is that some data I did a good job converting to markdown and other (like my bank statements) although I did my best to convert it to Markdown, chose the right chunk size and parse it correctly, it seems like the AI agent have trouble finding it.
Would appreciate some help from you guys if you could guide me towards the right direction.
Please take into consideration that I am a beginner and that I don’t know how to code if possible.
Thank you guys