A policy-writing coach and Unit 4's voice + defense partner: used in Blocks 9 (recommendations drafting), 10–11 (Brief drafting + Panel rehearsal), and 12 (Panel day). Two jobs: it coaches your Chennai Climate Brief into the voice an actual Greater Chennai Corporation official expects, direct, audience-aware, recommendation-anchored, no academic-essay creep, and it role-plays the tough panel questions before the live defense: cost, feasibility, equity, science credibility. It treats policy-brief voice as a learnable skill, not innate. For the science underneath, it hands you to the Climate Modeler.
Launch The Policy Aide: links coming soon. Ask Ms. Jayanthi/Mr. Ignash for access.How to use this partner
Start with your reader, not your draft: the Policy Aide's first question is always "who's your reader, and what do you want them to do after reading?" Then paste a section and it suggests specific edits rather than rewriting for you. For Recommendations it pushes for who-acts, when, what it costs, who benefits, who's burdened. Before the Panel, ask it to play a tough GCC official.
Two prompts you can paste in to get started:
- "Here's my Executive Summary draft. Tell me where it reads as academic essay instead of policy brief, and what specifically to change."
- "Play a skeptical GCC official. Ask me the hardest cost and feasibility questions about my top recommendation, one at a time."
Opening message
"Hi: I'm the Policy Aide. What part of your Climate Brief are you working on? Before we draft, tell me: who's your reader (a Greater Chennai Corporation official?), and what do you want them to do after reading? Audience focus is the whole game here."
Example exchange: what working looks like
| You | The Policy Aide |
|---|---|
| Here's my Executive Summary draft. "The science is clear that climate change poses significant challenges to Chennai requiring urgent action." | Honest feedback: that sentence is academic-essay voice. "The science is clear," "significant challenges," "urgent action": all general, all hedged. A policy-brief Executive Summary should be specific. Try this revision: "Chennai faces three concrete climate impacts in the next 25 years: [name them]. This brief recommends [name the top action]. The cost of inaction is [specific consequence]." Now: what are YOUR three concrete impacts and your top action? Let's build from those. |
Watch out for
Pressed hard, the Policy Aide can drift into writing the Brief for you, and words you didn't write are words you can't defend under Panel questioning. If you notice it generating your draft instead of editing it, pull back: "suggest edits to MY draft, don't write it." It should redirect to what evidence you have and what action you want the reader to take. Recommendations it proposes that aren't grounded in your evidence are the same problem: the policy reasoning has to be yours.
It can also get so focused on voice that it stops noticing whether the underlying science holds. The Policy Aide doesn't check science, that's the Climate Modeler's job. Anything quantitative that survives into your Brief still gets verified against the IPCC or your cited sources, whichever partner it passed through.
Before you go: documentation
At the end of a substantive Policy Aide conversation, the partner will remind you:
"Before you go: note in your AI Documentation: which voice problem did our conversation help you fix, in which section? And what's one suggestion I made where you decided to take a different angle, and why? That second entry is the evidence that the policy reasoning is yours, not mine."