Vibe coder here.
I built a lot of complicated proof of concepts before I started working with Agents and Agno. I have solved a million other problems. But, I can’t solve this, what seems to be the simplest problem of them all.
I have 3 software products; let’s call them X, Y and Z.
All products have features; let’s call them A, B and C.
I have 6 RAG databases, each software product has it’s own dedicated RAG databases.
This is what I need:
User: Tell me about A
Agno: Do you want to know about A in X, Y or Z?
User: X
AGNO takes “A” adds it to “X” so the full query becomes “A X”.
AGNO sends “A X” to the RAG.
I cannot get Agno to add the first query to the second query.
I cannot get Agno to add “A” to “X”
I cannot get Agno to create the combined query “A X”.
Why?
I have tried this a million different ways - with prompts, with slot filling, with hardcoded scripts.
But, it just won’t do it in any deterministic, repeatable way.
Sometimes it works. But just sometimes. Like, 10%-20% of the time.
Why is this so difficult?
For sure, this kind of query combining “slot-filling” must be a massive part of enterprise LLM implementation?
Is this not something Agno is designed for?
Please point me at documentation, repos, code examples, anything that will set me on the right path.
From all my research Agno seems like the best multi-agent architecture available, but I am lost at this point and can’t see a way forward. I cannot build anything else with Agno without first being able to clarify the users full-intent as part of the initial agent conversation flow.
Thanks for showing me the light!