Can an agent find, buy, and recommend your content?

Model the full machine journey. See where it breaks, why it breaks, and which metadata or policy change removes the friction. Results are scenario estimates, never claims of live indexing.

Content profile

Choose a baseline, then make it yours.

Content type

Access model

Public title

Agent-facing summaryState what it contains, who it helps, and the expected output.

Audience

License

Price per use

Currency

Access method

Freshness

Canonical page or endpoint

Payment rail

Provenance signal

Machine-readable signals

0readiness

Machine journey needs work

Complete the profile to calculate where an autonomous agent is likely to encounter friction.

This diagnostic evaluates supplied signals. It does not crawl the URL, confirm indexing, or execute a payment.

Follow the agent, decision by decision.

Run the simulated journey one checkpoint at a time. Each step names the signal used, the current decision, and the most useful recovery action when the path fails.

Agent journey

Awaiting analysis

Discovery

The agent starts from a public listing, search index, or known catalog.

Your content

Research agent

Buying agent

Teaching agent

Planning agent

Signal inspected: Catalog presence
Current result: Not run
Recovery: Analyze profile
Confidence: —

Prove what changed.

Save a baseline, improve the weak signals, then compare the score and the exact machine-readable capabilities gained.

Before

Saved baseline

—

Current

Live profile

Make the next agent decision easy to explain.

Export a compact readiness report with the assumptions, scores, generated metadata, and prioritized fixes.

Content readiness: declared signals and a simulated journey

Your browser does not support video.

Read the explanation

The free-article preset supplies declared signals, not a crawled page. Discovery seventy five, evaluation one hundred, transaction one hundred, retrieval eighty, trust one hundred and recommendation one hundred average ninety two point five, rounded ninety three. Turning off schema subtracts twenty discovery, twenty five evaluation and fifteen recommendation points, yielding eighty three overall. Bars encode ninety three and eighty three at four pixels per index point. This ten-point difference is fixed rubric arithmetic rather than observed agent behavior or search visibility. No supplied endpoint is requested. Saving a baseline stores a copy of the score and subscores in page memory; it is not a durable audit history or external evidence. Discovery assigns thirty five points to crawlability, twenty five to catalog, twenty to llms text and twenty to schema. These are unchecked form declarations. Bars compare eight journey steps with six score dimensions at forty pixels per count. A step labeled pass requires its dimension score at least sixty; confidence is merely that score floored at twenty five percent, not model uncertainty. Running the journey advances every step on a timer even if an earlier one displays Pause, so friction does not stop execution. With no actual retrieval, payment, citation or recommendation, the journey is an educational state display. Native input, save-baseline, next-step and actual text-report export preserve this distinction locally. A nonempty endpoint grants thirty five retrieval points. Retrieval method adds twenty five unless contact request, which adds five. Bars encode the thirty five endpoint contribution and five contact-method contribution at five pixels per index point. Validation only checks a title, nonnegative price and an HTTPS prefix; it does not verify a reachable endpoint or schema semantics. Invalid form makes composite zero while component scores remain computed. Generated JSON-LD puts update-frequency words into dateModified rather than an actual date, so output is not independently validated metadata. The actual exported text correctly states it does not confirm live indexing, retrieve URLs or execute payment. This narration corrects the old validated-JSON-LD implication. Local four-width original drawing and all-footer placement and playback are subset checks, without clipboard-content or public evidence.