InvestorsCareersClient support
TulanFabric + Agentic AI
Public Sector

A data agent that answers caseworker questions in seconds

A state agency grounded a Fabric data agent in certified program data, giving caseworkers and analysts instant, governed answers.

Fabric Data AgentsSemantic ModelsAzure OpenAIPurviewLoom Ledger

The challenge

Program analysts spent most of their week answering ad hoc data requests from caseworkers and legislators. Data lived across eligibility, case management, and payment systems with strict privacy rules.

What we did

Tulan ran an AI value workshop that ranked 60 use cases, then built certified semantic models for eligibility, enrollment, and outcomes. A Fabric data agent grounded in those models answered natural-language questions with row-level security intact. An evaluation harness scored accuracy weekly, and Loom Ledger tracked token spend per question.

5xfaster time-to-answer for analysts
92%evaluated answer accuracy
3,200caseworkers with access
-45%ad hoc data requests

The outcome

Caseworkers and analysts get governed answers in seconds. Ad hoc requests to the analytics team fell 45 percent, freeing analysts for program evaluation work.

“The agent respects every privacy rule we have and still answers in plain English. That combination is rare.”
Chief Data and Analytics Officer

Let's put agents to work on your Fabric estate.

Book a 45-minute working session with a Tulan architect. Bring your hardest data problem; leave with a plan.