Initial commit - Route Commerce platform

src/app/api/ai/report-explainer/route.ts

@@ -0,0 +1,87 @@
+import { NextRequest, NextResponse } from "next/server";
+import { getAdminUser } from "@/lib/admin-permissions";
+import { getAIClient } from "@/actions/integrations/ai-providers";
+
+const REPORT_PROMPTS: Record<string, string> = {
+  "orders-by-stop": `This is an Orders by Stop report. Each row has: stop_name, city, state, date, order_count, gross_sales, pending_count, completed_count. Identify top-performing stops, underperforming stops, and patterns in pending vs completed orders.`,
+  "sales-by-product": `This is a Sales by Product report. Each row has: product_name, units_sold, gross_revenue, avg_price. Identify top sellers, underperformers, and pricing anomalies. Flag any products with unusually high/low average prices.`,
+  "fulfillment": `This is a Fulfillment Breakdown report. Each row has: fulfillment_type (pickup/shipping), order_count, revenue, pct_of_total. Analyze pickup vs shipping split and revenue distribution.`,
+  "pickup-status": `This is a Pickup Status report. Each row has: stop_name, city, date, total_orders, pending, completed, canceled. Identify stops with high pending rates, high cancellation rates, and completion speed.`,
+  "contact-growth": `This is a Contact Growth report. Each row has: date, new_contacts, imports, total. Identify growth trends, import spikes, and rate-of-change patterns.`,
+  "campaigns": `This is a Campaign Activity report. Each row has: campaign_name, status, campaign_type, sent_at, messages_logged. Identify which campaign types perform best, timing patterns, and engagement levels.`,
+};
+
+const SYSTEM = `You are a data analyst for a B2B produce wholesale distribution platform.
+You analyze report data and return structured insights in plain English.
+Always be specific and cite numbers from the data.
+Format your response as JSON with this exact shape:
+{
+  "summary": "2-3 sentence plain-English summary of what this report shows",
+  "keyInsights": ["insight 1", "insight 2", "insight 3"],
+  "suggestedActions": ["action 1", "action 2"]
+}
+Be direct and actionable. Do not use hedging language.`;
+
+export async function POST(req: NextRequest) {
+  const adminUser = await getAdminUser();
+  if (!adminUser) {
+    return NextResponse.json({ error: "Unauthorized" }, { status: 401 });
+  }
+  if (!adminUser.can_manage_reports) {
+    return NextResponse.json({ error: "Forbidden" }, { status: 403 });
+  }
+
+  try {
+    const { reportType, dateRange, brandId, reportData } = await req.json();
+
+    if (!reportType || !reportData) {
+      return NextResponse.json({ error: "Missing reportType or reportData" }, { status: 400 });
+    }
+
+    // Brand scoping: platform_admin passes explicit brandId; brand_admin limited to own brand
+    const effectiveBrandId = adminUser.brand_id ?? brandId ?? null;
+    if (!effectiveBrandId) {
+      return NextResponse.json({ error: "Brand ID required" }, { status: 400 });
+    }
+
+    const aiResult = await getAIClient(effectiveBrandId);
+    if (!aiResult.client || !("client" in aiResult)) {
+      return NextResponse.json({ error: "error" in aiResult ? (aiResult as { error?: string }).error ?? "AI not configured" : "AI not configured" }, { status: 503 });
+    }
+
+    const prompt = REPORT_PROMPTS[reportType] ?? `This is a ${reportType} report. Analyze it.`;
+
+    // Build a compact sample of rows (max 20 for token efficiency)
+    const sampleRows = Array.isArray(reportData) ? reportData.slice(0, 20) : [];
+
+    const userMessage = `Date range: ${dateRange?.start ?? "unknown"} to ${dateRange?.end ?? "unknown"}
+Brand: ${effectiveBrandId}
+${prompt}
+
+Data:
+${JSON.stringify(sampleRows, null, 2)}
+
+Return JSON with summary, keyInsights (array), and suggestedActions (array).`;
+
+    // eslint-disable-next-line @typescript-eslint/no-explicit-any
+    const client = aiResult.client as { chat: { completions: { create: (opts: unknown) => Promise<{ choices: Array<{ message: { content: string } }> }> } } };
+    const response = await client.chat.completions.create({
+      model: aiResult.model,
+      messages: [
+        { role: "system", content: SYSTEM },
+        { role: "user", content: userMessage },
+      ],
+      response_format: { type: "json_object" },
+      temperature: 0.3,
+    });
+
+    const content = response.choices[0]?.message?.content;
+    if (!content) {
+      return NextResponse.json({ error: "No response from AI" }, { status: 500 });
+    }
+
+    return NextResponse.json(JSON.parse(content));
+  } catch (err) {
+    return NextResponse.json({ error: "AI analysis failed" }, { status: 500 });
+  }
+}