Policy Analytics Agent · Use case
How can policy teams question their own data with an AI agent on their own infrastructure?
A policy analytics agent lets policy staff ask questions of the agency's own data in plain language. It writes the query, runs it with read-only access, and shows the result with the query and the assumptions behind any scenario. An analyst checks the method before a figure is used in advice. The agent runs on infrastructure the agency controls.
A question for the data team.
And a wait for the answer.
- ✗ 01Policy staff wait for the data team to answer each question, so fewer options are tested.
- ✗ 02Programme data is sensitive, and a forecast with hidden assumptions cannot support a decision.
Where it stops for a person
Before a figure is used in advice.A scenario or forecast is an estimate. An analyst checks the method before a figure goes into advice or a publication.
It answers from your data.
Analysts check the method.
Step · 01
The programme data is connected with read-only access.
Step · 02
A policy officer asks a question or describes a scenario.
Step · 03
The agent writes the query and runs it.
Step · 04
It shows the result, the query and the assumptions it made.
Step · 05
An analyst checks the method before the figure is used.
What you set up
- Connect the data and describe what each table holds.
- Set who may ask about which data.
- Write test questions with known answers.
Where it runs
The agent, its knowledge store and the model run on infrastructure you own or control. You choose the model and serve it yourself, so questions, documents and answers are not sent to a public AI service.
How you get it today
This page is labelled Use case: the agent is built in Katonic Studio. You describe the job, connect what is named above, add test questions and release the agent once it passes them. When the Catalog holds a tested template for this job, this page will carry the Template label and show how to take it.
Can it change the data?
No. Its access is read-only.
How do we check a forecast?
The agent shows its query and assumptions with every result, for an analyst to check.
Where is the data processed?
On infrastructure the agency controls, with a model it serves.