[](/) Case study · Build · Speed · Founder judgment # AgentSite. "AEO is the new SEO." Idea to first paying customer in four weeks. We kept seeing the same gap across client work: founders shipping AI-native SPAs that ChatGPT, Claude, and Perplexity literally couldn't read. Empty `<div id="app">` — no citations, no answers, no traffic from the answer engines. So we built the fix. Four weeks from kickoff to first paid customer. Now serving production tenants, including wilde.agency itself. The engagement 4 weeks idea → first paying customer 7+ production tenants 13,250 requests on wilde.agency alone 53% snippet hit rate ## Build the tool we wished existed — fast enough to validate the need before it commoditized. The answer engines (ChatGPT, Claude, Perplexity) had quietly become a top‑of‑funnel channel that almost no SPA could appear in. Server‑rendered sites won citations. Vite/Vue/React SPAs got skipped. Every founder we advised was about to discover this the hard way. Mandate: ship a working product fast enough to be the answer, not the post‑mortem. ## Ship in four weeks. Iterate on what only paying customers tell you. Week one: the AEO scoring engine — five layers (crawler access, AI‑readable content map, structured data, content quality, external measurement) that anyone could run on any URL in under three minutes. Public score report URL was the whole top‑of‑funnel. Week four: first paying customer. The product was just enough — public score, snippet generation, snippet hosting, cache. No payment tiers yet. No dashboard. Post‑launch iteration was driven entirely by what paying customers asked for: usage tracking → payment tiers (Studio plan at $0.01/unit, weekly allowance pools) → per‑tenant AEO scoring history → page‑level inspection with live telemetry → bot‑policy toggles for /robots.txt allow‑listing. wilde.agency is itself a paying tenant. We score 42/100 in front of you. Telemetry is real (13,250 requests, 5,624 cache hits, 53% hit rate at the time of writing). When we say we dogfood our own tooling, that's what we mean. ## What we shipped [](/work/agentsite/landing.png "Open full-size") The wedge: ChatGPT, Claude, and Perplexity don't run JavaScript — they see your SPA as an empty. [](/work/agentsite/score-report.png "Open full-size") Public AEO score report — 0–100 across five layers. Anyone can run it on their own URL in under three minutes. [](/work/agentsite/dashboard.png "Open full-size") Multi‑tenant dashboard with unit balance, recent reports, and live tenant scoring. [](/work/agentsite/snippet-health.png "Open full-size") Per‑tenant snippet health. Generated files (/llms.txt, /llms-full.txt, /sitemap.xml, /robots.txt) served when an AI bot requests the page. [](/work/agentsite/page-inspection.png "Open full-size") Page‑level inspection — requests, scans, generates, last‑seen per path. ## What it taught us ### Experience sees beyond the obvious. AEO didn't exist as a category when we started. Founders thought their AI bot traffic was a quirk of their analytics. Sitting across dozens of advisory and build conversations, the pattern was unmissable. The build wasn't the hard part — recognizing it as a build at all was. That's what experience does in an information‑abundance era: it cuts through diluted signals to name the integration problem that's always been there. The gap turns into a product. The product turns into a category. Live at [agentsite.app](https://agentsite.app). ## Have a gap you want named — and shipped? Operator judgment, a foundation that ships, and the team to make it real before the market catches up. [Book an architecture review →](/start?utm_campaign=case-agentsite) Got a question?