Full AI Search Audit
One URL in. Six GPT-4o-mini agents out — Technical SEO, Content SEO, auto-fixes, AIO, AEO, and GEO — emailed as a Markdown report.
Problem
SEO, Google AI Overviews, featured snippets, and ChatGPT/Perplexity citations are four different jobs. Agencies still do them as four retainers.
What we built
One scrape. Six agents. One report. Technical SEO finds code-level gaps. Content SEO scores keywords and readability. SEO Fix Generator returns JSON (meta, H1/H2, alt, schema). AIO scores Google AI Overviews. AEO scores featured snippets and voice. GEO scores citability in ChatGPT, Claude, and Perplexity.
How it runs in real time
In production we treat this as a 10-minute intake. Paste taskwagon.com (or any client URL) into the n8n form. While you stay on the call, scrape → six agents → Merge All → Aggregator → Format Full Report → Google Doc + Gmail. The operator does not rewrite the audit; they send the Doc link. Same-day use: ship the JSON block to the Full Stack squad for metadata and schema on the Next.js site.
Pipeline
01
Form + scrape
Operator submits the URL. n8n pulls the live HTML so agents see what Google and AI crawlers see — not a screenshot of the SPA.
02
Six agents in parallel
Technical SEO, Content SEO, SEO Fix Generator, AIO (AI Overviews), AEO (snippets + voice), GEO (ChatGPT / Claude / Perplexity citations).
03
Merge + format
Merge All + Aggregator stitch sections 1–6 into one Markdown pack: critical issues, quick wins, opportunities, JSON fixes.
04
Deliver
Google Docs document is created. Gmail sends the report. WhatsApp is wired but off until the client wants chat delivery.
In the canvas
- ▸ n8n Form Trigger — single URL field, no custom app
- ▸ Live HTML scrape of the submitted landing page
- ▸ Six parallel OpenAI agents (not one mega-prompt)
- ▸ Merge + Markdown formatter → Google Docs
- ▸ Gmail delivery; WhatsApp notify node on the canvas (deactivated)
This is a live n8n engine, not a slide deck. A client (or our own product team) pastes a landing URL into an n8n form. The workflow scrapes the live HTML, fans out to six specialised GPT-4o-mini agents in parallel, merges their findings, formats a full Markdown report, writes it to Google Docs, and emails it on Gmail. WhatsApp delivery exists on the canvas and is toggled off until a campaign needs it. We ran it against TaskWagon and got a real scorecard: missing meta description and H1, AIO/AEO/GEO readiness at 3/10, plus ready-to-paste JSON for title, description, headings, alt text, and LocalBusiness schema.
Proof
Results
- ▸ Real TaskWagon run: missing meta + H1 called out with copy-paste fixes
- ▸ AIO / AEO / GEO readiness scored (3) with FAQ and snippet rewrites
- ▸ Auto JSON: meta_title, meta_description, schema LocalBusiness