Skip to main contentSEO for AI Agents.
How it works

URL in.
Plan out.

No black box. Every score in SEO for AI Agents traces to a deterministic check, evidence captured at audit time, and a weight chosen with intent.

~90s
Free-tier end-to-end runtime, including AI brief generation.
— Median wall-clock, May 2026
5
Pipeline steps, every one inspectable in the receipts drawer.
— SEO for AI Agents architecture
2-pass
AI brief is drafted, then critiqued against a rubric, then revised before you read it.
— Gemini 2.5 Pro orchestration
Typical SEO tool
This audit
Input
Spreadsheet of keywords + GSC export.
Input
One URL. We discover the rest.
Rendering
Plain HTML fetch — SPAs come back empty.
Rendering
Chromium-equivalent JS rendering on every page.
Brief
Templated text wrapped around a chart.
Brief
AI-drafted, self-critiqued, owner-assigned 30/60/90 plan.
01

Drop in a URL

Any public URL works. We auto-discover up to 500 pages from sitemap.xml, llms.txt, and same-origin links - capped per tier so the run finishes in 90 seconds on Free and a few minutes on Agency.

02

JS-rendered crawl

Every page is fetched through a Chromium-equivalent renderer. We capture status codes, response headers, rendered HTML, parsed markdown, JSON-LD blocks, links, images, and Core Web Vitals.

03

45 deterministic checks

Across Foundational, Technical, On-Page, AI Overviews, GEO, Local, and E-E-A-T. Each finding carries the raw evidence - the offending header, the JSON-LD object, the URL list, the heading text. You can audit our audit.

04

Weighted scoring

Each check has a weight from 1 to 6. Dimension scores are weight-normalized against the maximum achievable; the overall is the weight-normalized average across all non-N/A checks. No vibes.

05

AI brief - drafted, critiqued, shipped

Gemini 2.5 Pro writes a one-line verdict, a CEO brief, a technical brief, and a 30/60/90 plan with owners. A second pass criticizes that draft against a rubric: is the headline metric counter-intuitive? Is every recommendation actionable? The revised draft is what you read.

Run a free auditDeep methodology