Readiness
—
scored public-surface checks
AgentFit dashboard
Agent readiness
Ensure customers can reach, understand, and complete work through your site with AI agents.
—
Open your dashboard
Create an account or use your existing account token. Tokens are shown once and only their hashes are stored.
Add a site to begin
Diagnostics
BetaNo diagnostic result yet
Run the seven-stage audit to measure the site without treating missing instruments as zero.
- URL
- —
- HTTP
- —
- Measured
- —
idle
Agent-market fit
Latest runCompletion
—
attempted tasks completed
Mention rate
—
answers naming the site
Citation rate
—
answers citing the site
Prominence
—
share and position in answers
See how often AI assistants cite your site
How prominently it features, and what to improve across OpenAI, Anthropic, and more. No completed run selected. A dash means not measured, never zero.
Run ledger 7 stages
Each stage writes one row when it finishes. Unavailable instruments are named explicitly and remain distinct from failures and zero values.
| Stage | Status | Finding | Unavailable instrument | Evidence |
|---|
Source: the selected run’s persisted stage rows and R2 evidence keys.
Content-hash verification
AI playground
BetaExperiment in the AI Playground
Compare prompts and responses across models, and see how AI agents interpret your content before you ship changes.
Learn more about agent readiness
Read blogAs web traffic shifts from human visitors to autonomous AI engines, structuring your site for machine consumption is the next frontier of web development: from SEO to answer-engine optimization (AEO), so agents can discover, digest, and cite your content. Agent readiness is still emerging, and the standards behind these checks keep evolving.
Stage 01
Map the public surface
Read the site the way an agent does: crawler policy, sitemaps, Markdown, entity data, and agent interfaces.
origin checks + passive company-surface discovery via crt.name · no key required
Commerce readiness
informationalStage 03
Ask grounded answer engines
Compare exact citation, mention, prominence, and share-of-voice measurements across enabled providers.
Prompt flow
intent tiersCitation overlap
hostsStage 05
Diagnose the answer path
Readiness and recommendation are separate conditions. Citation and mention booleans remain exact; models add diagnosis, not scores.
requires a completed map + answer panel
Operator readiness
ORICompletion funnel
per taskFindable · citable · completable
joinAgent-ready dimensions
paper joinOptimization
Cover prompts the site lost
Ingest source passages, retrieve supporting evidence, and produce capsules only when their claims already exist in the source.
uses the latest answer panel; gate-passed capsules only
Stage 06
Build the executable path
Turn deterministic gaps and observed provider behavior into generated, source-bound files rather than advice.
Run Map first.
Stage 07
Ship and measure
Import content, approve source-led edits, probe citations, and measure interventions against the matching control period.
discovers sitemap.xml or RSS automatically · re-imports update, never duplicate
scores every post and drafts surgical fixes for the highest-opportunity items
checks whether the selected post is cited
Approved edits receive a before snapshot and 28/56-day readings. A post-level change is reported only against blog-wide movement for the same dates.
Operator evidence
Operator traffic
Import access logs or operator summaries to separate crawling from customer referrals. The crawl-to-refer ratio is pages fetched per visitor sent back. Access logs cannot measure referred visits, so that ratio stays a dash, not a zero.
current site · latest imported window
Just handle it
Agent playground
Send one customer prompt through the answer panel. Transactional prompts also attempt the site's real task path within the selected safety boundary.
Provider access
Connect answer engines
Keys remain in this browser’s localStorage and are sent only with the corresponding probe request. The server does not persist them.