The problem
Leadership questions about the business ("why is churn up in this segment, and what do we do?") cross every table: customers, deals, tickets, usage, invoices. An LLM can hold the conversation but cannot compute the answer, and a SQL bot can count rows but cannot say which action moves the number.
How it works
The agent's tools are Aito queries. A KPI snapshot is a set of _predict calls; optimising a KPI chains _predict with $why, _relate for the drivers, and _recommend for the lever, then predicts the lift. Customer lookups and examples are _query, and revenue estimates are _estimate. The LLM chooses the tools and writes the answer.
Acting is separate from answering: the agent can draft a play, but launching it waits for a person to approve.
For the full architecture, see the technology overview. For the broader narrative across multiple use cases, read The Predictive Application.
See it live
This use case runs in the 🤖 Agent demo today. Click through to the live application and inspect the queries that produce the result.
