ELI RESEARCH

AI-agent action readiness: from readable website to permissioned action

A website can be excellent to read and still expose no safe write actions.

Readable → structured → invokable → authorized → observable
Research in progress

Action-readiness benchmark not yet released. No percentages or distributions are published.

RELEASE STANDARD

Methodology is publishable before findings are.

Eli can publish the question universe, schema, definitions and quality controls before a benchmark release. It will only publish an aggregate when the underlying observations are complete, reviewable and described with their coverage and limitations. Until then, this page makes no performance, ranking or customer-outcome claim.

Maturity and categories

Assess readable, structured, invokable, authorized and observable; categories include read, calculate, configure, book, request, create, update and purchase, each with different risk.

Audit dimensions

Check name/purpose, inputs, validation, authentication, authorization, confirmation, failure state and audit/result.

Future findings

Release actions per site, maturity distribution, missing controls, category differences and risk availability only with audited actions.

FAQ

Questions about the research design and release boundary.

Action readiness?

Safe explicit business actions for agents.

Form action?

It can be.

MCP increases?

Only with correct controls.

High scores autonomy?

No.

Purchase goal?

Not necessarily.

CONNECT YOUR WEBSITE

Assess your action readiness

AI-agent action readiness measures whether a website exposes useful, clearly defined business operations with inputs, validation, authentication, authorization, confirmation and observable results.

Assess your action readiness