Agent Surface

FreeNo loginAbout 20 seconds

See what AI agents
can do with your website.

A clear answer from real evidence.

A simple answer, not a technical audit

Know where you stand—and what to do next.

431 unique websites scanned440 completed scansSee the live Observatory

  1. 01We look

    We check the public pages an AI agent can reach.

  2. 02We explain

    You get a plain-English position on the Agent Surface Ladder.

  3. 03You decide

    See the improvements and agent tools most worth building next.

Not just another website score

One scan turns a website into an agent-readiness map

See what agents can read, answer and do; compare it with the live corpus; then get the first tools worth exposing — with the source evidence beside every recommendation.

Live benchmark

See where the market actually sits

Every completed site is placed on the same published five-rung rubric. The distribution updates as the corpus grows.

89
Invisible
304
Readable
25
Answerable
12
Callable
0
Transactable
Explore the Observatory →
AI crawler access

See which AI products can reach the site

SITEChatGPT: explicitly allowed✓GPTBot: allowed by default•Claude: explicitly allowed✓ClaudeBot: blocked×Gemini: allowed by default•Perplexity: explicitly allowed✓ChatGPTGPTBotClaudeClaudeBotGeminiPerplexity

Example radial access map — every product is equally weighted; its dot shows the site's published treatment.

Agent tool blueprint

Discover what your site could let agents do

We map observed forms, booking paths, pricing and content into concrete tool ideas. No LLM guesswork: each proposal is generated by rules from evidence the scanner found.

Two recommendations are revealed in the free report. Contact us to map the capabilities behind your public pages.

Agent access

Separate AI search from model training

We inspect six published controls and explain what each one governs, instead of treating every AI bot as the same visitor.

ProductSearchAI use
ChatGPTOAI-SearchBotGPTBot
ClaudeClaude-SearchBotClaudeBot
GeminiGoogle Search rulesGoogle-Extended
PerplexityPerplexityBotNot for training
Verifiable by design

Follow every claim back to what the agent saw

01 Public URL→02 Observation→03 Rubric score→04 Recommended move

Exact source URL, timestamp and safe evidence snippet sit beside every result. If the scanner cannot measure something, it says unmeasured — never zero.

A report worth keeping

Your complete findings, delivered by email

The public verdict shows where the site sits. The email gate unlocks the full evidence, benchmark, recommendations and copyable implementation prompt — for humans and agents alike.

Scan your website