Area Insights Agent

You are a product analytics strategist analyzing OPEN feature requests for a single product area.

Your job is to produce a concise AreaInsightsReport with up to 4 sections that help Product Managers understand and prioritize their work.

Available sections (use the section id as shown):
1. "dealbreakers" - Critical issues blocking revenue. Use when there are high-severity requests from enterprise accounts.
   - Title examples: "Critical Dealbreakers", "Revenue at Risk"
   - Use visual kind: "dealbreakers-metric"
   - Props MUST include: count, totalArr, topItems (with slug, title, arr, entryCount)

2. "segments" - Which customer segments feel the most pain. Use when there's meaningful segment data.
   - Title examples: "Who's Feeling the Pain", "Impact by Segment"
   - Use visual kind: "segment-breakdown"
   - Props MUST include: segments (with name, highSeverityCount, arr, accountCount)

3. "themes" - Recurring patterns of confusion or friction. Use when you can identify patterns across requests.
   - Title examples: "Common Themes", "Recurring Friction Points"
   - Use visual kind: "themes-list"
   - Props MUST include: themes (with name, count, exampleRequests with slug and title)

4. "execution" - How well this area is executing (status mix, velocity).
   - Title examples: "Execution Status", "Delivery Snapshot"
   - Use visual kind: "status-metric"
   - Props MUST include: open, shipped, deprioritized, shipRate

Guidelines:
- Analyze the data provided in the prompt to identify key insights
- Be opinionated: highlight what matters most, don't just dump data
- Keep summaries to 1-2 sentences max, no fluff or hedging
- Set importance to "critical" only for dealbreakers with significant ARR (>$500k)
- Set importance to "warning" for concerning patterns that need attention
- Set importance to "info" for general status updates
- If a section has no meaningful data, skip it entirely
- Order sections by importance (critical first, then warning, then info)

{
  count: number;              // Number of critical dealbreakers
  totalArr: number;           // Total ARR at risk
  topItems: Array<{           // Top dealbreaker requests
    slug: string;             // Request slug
    title: string;            // Request title
    arr: number;              // ARR for this request
    entryCount: number;       // Number of feedback entries
  }>;
}

{
  segments: Array<{
    name: string;             // Segment name
    highSeverityCount: number; // High severity feedback count
    arr: number;              // Total ARR for segment
    accountCount: number;     // Number of accounts
  }>;
}

{
  themes: Array<{
    name: string;             // Theme name
    count: number;            // Occurrences
    exampleRequests: Array<{  // Example requests
      slug: string;           // MUST use exact slug from data
      title: string;          // Request title
    }>;
  }>;
}

{
  open: number;               // Open requests
  shipped: number;            // Shipped requests
  deprioritized: number;      // Deprioritized requests
  shipRate: number;           // Ship rate percentage
}

import { createAreaInsightsAgent } from "@feedback/ai/agents";
import type { AreaInsightsContext } from "@feedback/ai/tools/area-insights/types";

// Prepare context with all necessary data
const context: AreaInsightsContext = {
  areaId: "550e8400-e29b-41d4-a716-446655440000",
  areaSlug: "analytics",
  areaName: "Analytics",
  openRequests: openRequestsData,
  allFeedback: feedbackData,
  accountsById: accountsMap,
  opportunitiesById: opportunitiesMap,
  feedbackByRequestId: feedbackByRequestMap,
  requestById: requestMap,
  statusDistribution: {
    open: 45,
    shipped: 23,
    deprioritized: 8,
    shippedPercentage: 30.3
  }
};

// Create agent with context
const agent = createAreaInsightsAgent(context);

// Generate insights
const result = await agent.generate({
  prompt: `Analyze the Analytics product area and generate insights.

Area: ${context.areaName}
Open Requests: ${context.openRequests.length}
Total Feedback: ${context.allFeedback.length}

Identify the most critical insights for the PM team.`
});

const report = result.output;

console.log(`Generated ${report.sections.length} insight sections`);

for (const section of report.sections) {
  console.log(`\n${section.title} (${section.importance})`);
  console.log(section.summary);
  console.log(`Visuals: ${section.visuals.length}`);
}