FireGEO
Real-time wildfire geospatial monitoring shipped as a paying product in 7 weeks
The Problem
A climate data startup had been processing satellite imagery feeds for two years, building a rich dataset of historical wildfire perimeters, spread velocity, and atmospheric conditions. They had the data science side solved. They did not have a product.
They’d tried building a dashboard in-house — it stalled at the “how do we charge for this” problem. They needed a working product with billing, user accounts, and a live data view in six weeks. The technical constraint was that their analysis pipeline could take 30–90 seconds per query; the UI had to feel responsive regardless.
What We Built
FireGEO extended an existing open-source wildfire monitoring scaffold into a full SaaS product. The key additions:
- SSE streaming layer — analysis results stream to the client progressively as they’re computed, so users see partial results within 2 seconds even if the full computation takes 90 seconds
- Credit-debit billing — credits are reserved at request start, deducted on completion, refunded on error; Stripe webhooks reconcile account balances
- JSONB result storage — Drizzle ORM stores heterogeneous analysis outputs (perimeter polygons, spread vectors, risk scores) as typed JSONB, indexed for fast retrieval
How It Works
The SSE route handler is the core piece. It reserves credits upfront, streams analysis chunks, then either commits or refunds:
export async function GET(req: Request) { const { searchParams } = new URL(req.url); const query = searchParams.get("q"); const userId = await getUserId(req); // Reserve credits before starting (never let analysis run unpaid) const reservation = await reserveCredits(userId, ANALYSIS_COST); if (!reservation.ok) { return new Response("Insufficient credits", { status: 402 }); } const stream = new TransformStream(); const writer = stream.writable.getWriter(); const encoder = new TextEncoder(); const send = (data: object) => { writer.write(encoder.encode(`data: ${JSON.stringify(data)}\n\n`)); }; // Run analysis in background, stream results runFireAnalysis(query) .on("chunk", (chunk) => send({ type: "chunk", data: chunk })) .on("complete", async (result) => { await commitCredits(reservation.id); await storeResult(userId, query, result); send({ type: "complete", result }); writer.close(); }) .on("error", async (err) => { await refundCredits(reservation.id); send({ type: "error", message: err.message }); writer.close(); }); return new Response(stream.readable, { headers: { "Content-Type": "text/event-stream", "Cache-Control": "no-cache", Connection: "keep-alive", }, }); } The Outcome
The credit-reserve-then-commit model meant zero revenue risk from partial or errored analyses. The SSE streaming meant the product felt fast even on a 90-second analysis — users saw fire perimeter outlines populating the map in real time, which became the product’s defining demo moment.
The startup used the billing architecture as a template for a second product line six months later.