AI agent that retrieves data from databases

I have a business idea where AI will retrieve required data from database (non-SQL) to address user query.

It's unfeasible to "teach" AI specific query language, so I am reverting to create a set of custom tools-functions that does this job for specific task.

Do we have any good examples or guidance on implementation of this idea?

Is langchain is the tool of choice for this orh there are better frameworks for this?

Sort of:

"Tell me what caused an anomaly in server traffic for the last 24 hours?"

LLM will run query to:

Find timeframe of anomaly (spike in events for the last 24 hours)
Run the query of source (region), IP address, browser type, etc... for that timeframe
Compare this data to typical average traffic during this time - run "calculate an average..."
Report resuts: "This IP address, from this region was trying to hit these pages 10x times..."

That's the logic.

Interested in possible implementations of this

1 comment