Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for policy analysts

Policy Brief Evidence Agent

A brief evidence section where every claim matches a rated source

ResearchCreatingGovernment
See how it worksGet access
Policy Brief Evidence Agent: what goes in, what the agent does and what you get

What it does

Policy briefs need to say what the evidence shows, how strong it is and what is still unknown, and analysts rarely have time to read every study. This agent takes the brief question, searches research databases, government reports and think tank studies, and rates each source by study type and relevance. It builds an evidence table and drafts the evidence section from the rated sources. Then it checks that every claim matches a quote or figure in its source; claims stronger than the source are toned down or removed and the section is redrafted. If fewer than three relevant sources turn up, it widens the search terms and dates. Weak sources can give context but not conclusions. The analyst writes the recommendations and approves the brief. Edge case: when studies disagree, the draft shows both sides with their strength rather than picking one.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Analyst enters the brief question 2 USES A TOOL Search research databases and government reports 3 DOES Rate each source by study type and relevance 4 DOES Draft the evidence section from rated sources 5 CHECKS THE RESULT Does every claim match a quote or figure in itssource? If not: tone down or remove the claim and redraft. Backto step 4. 6 CHECKS THE RESULT Are there at least three relevant sources? If not: widen search terms and dates. Back to step 2. 7 YOU APPROVE Analyst approves the brief 8 RESULT Brief with evidence table
Read the steps as a list
  1. Analyst enters the brief question
  2. Search research databases and government reports
  3. Rate each source by study type and relevance
  4. Draft the evidence section from rated sources
  5. Does every claim match a quote or figure in its source?If not: tone down or remove the claim and redraft. Back to step 4.
  6. Are there at least three relevant sources?If not: widen search terms and dates. Back to step 2.
  7. Analyst approves the briefThe agent waits here for your OK.
  8. Brief with evidence table

How it decides

It rates sources from strongest (trials, systematic reviews) to weakest (opinion), and only lets the brief state as fact what strong sources support.

  • Weak sources can support context, not conclusions
  • Disagreeing studies are both shown
  • Claims stronger than the source are rewritten

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Source types to include
  • Date range for studies (default last 10 years)
  • Citation style
  • Brief length

What keeps you in control

It always asks you first

  • Analyst approves the brief and writes recommendations

Hard limits

  • Never invents citations
  • Does not write policy recommendations

It stops when

  • Done: all claims verified
  • Stop: no relevant sources found after widening search

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensFor a brief on four-day school weeks, the agent found 9 studies. The draft said they 'improve attendance,' but only one quasi-experiment showed a small gain and two showed none. The claim check failed, so the agent rewrote the line as 'mixed evidence on attendance' with all three studies cited, then re-checked every claim. The analyst, Dana Ruiz, added recommendations and approved the brief on Sep 18.

More agents for policy analysts

Reporting · Research

Stakeholder Brief Update Agent

Current, documented stakeholder briefs.

8 steps1 check5 jobs
Reporting · Government

Campaign Promise Delivery Tracker Agent

A sourced status table for every promise that the official can publish with confidence

10 steps2 checks4 jobs
Creating · Automation

Amendment Comparison and Impact Agent

A short briefing note on what each amendment changes, with conflicts and knock-on effects listed and sourced

11 steps2 checks3 jobs
Analysis · Quality control

Fiscal Note Review Agent

A memo that shows whether the fiscal note's totals can be reproduced, which assumptions drive them and a range

10 steps2 checks3 jobs