Skip to content

See if an AI can buy from your store — free.

After you fix, you can measure once more.

We only measure stores you control.

This check stops us from running other people's shops. On BASE, paste one meta tag in HTML tag settings.

Add the free HTML Tag Manager app in your admin (Apps) and paste the one-line meta tag we send you. No custom domain needed — it works on your *.base.shop URL as-is.

Other ways to confirm

Agentrix drives a real browser through someone's store, so it never runs on a site whose control has not been proven. Immediately before each run we look for any one of three proofs, all of which need write access to the site or its DNS.

  1. DNS TXT record on your domain (or _agentrix.<your domain>): agentrix-site-verification=<token>
  2. A file at /.well-known/agentrix-site-verification.txt containing your token
  3. A <meta name="agentrix-site-verification" content="<token>"> tag on your home page

Nothing runs until one of the three checks out. The request waits and retries once a day; after 6 checks it is withdrawn. We email your token with the receipt, so you can set this up after you apply.

BASE: The HTML Tag Manager app (listed at no cost in the app store) injects a custom tag into the head of every page except cart and checkout. No custom domain is needed — it works on the *.base.shop default domain as-is (neither the app page nor the help conditions it on a domain). If you do use a custom domain you also control the DNS and can add a TXT record (BASE does not sell domains; because a CNAME is required, the store URL must be a subdomain such as shop.example.com).

BASE — Official help (the platform's own documentation)

Tick owner (or authorised-party) consent to send.

See public pages first, no signup

Study flow, what we never do, and what we publish

A sample of the report you receive

A real run on our own site. The agent was sent to a login page when it tried to sign up, and stopped at 26 points. The screen where it stopped is saved.

View the sample report

From sign-up to results

The sign-up form
Four fields — store URL, platform, your name, and email — plus a required checkbox confirming you are the owner (or an authorised party).
Confirming you run the store
This keeps anyone from diagnosing someone else's store. On BASE, Color Me Shop, Shopify and similar platforms it is one line pasted in your admin (no custom domain needed). We email that line after you apply, so you can do it afterward. We recheck once a day; after 6 checks with no proof, the request is withdrawn.
Time to results
Once ownership is verified you can run the diagnosis yourself from the console at /en/app. We also email the share link (as a guide, usually within two business days).

What you get

  • An agent-reachability report for your store, free
  • Where the run stopped between signup, browse, cart and pre-checkout — with barrier IDs
  • What to change on your platform (permission: theme / app / plan / platform)
  • One free retest after you fix, to see whether the score moved

What the diagnosis never does

  • Bypass a CAPTCHA — the run stops on detection, and that stop is the result
  • Complete a purchase or payment — runs always end at pre-checkout
  • Diagnose a store without a consent record

Collection progress

Wave 1 is now accepting participating stores (results to be published October 31, 2026).

First publication (if minima are met): 31 Oct 2026 · This wave accepts until: 15 Oct 2026

On 2026-09-06 we moved the signup deadline from 09-15 to 10-15. Signups were zero, but the cause is on our side: the only sample of what you receive is one failed run on someone else's store, and almost nobody has seen the page (12 visitors a week). This is not a market verdict.

September 2026 study aggregates (current state, with sample) →

Limits of this study

This is a volunteer sample, not a random draw of Japanese EC. Stores that care about agents self-select. Do not generalise a later “X% of BASE” figure past that.

Why participation is opt-in

We do not walk other people's stores without permission: a diagnosis that actually operates signup or the cart only runs when the site's operator asks for it, and a study is no exception. So this study runs entirely on stores whose owners volunteer them.

What is and is not published

Published

  • Per-platform aggregates (e.g. what share of BASE stores stop at signup) — after minimum N
  • Distribution of stop reasons (barrier IDs; CAPTCHA counted as Policy, separately)

Never published

  • Individual store names or URLs (unless the owner separately opts into Walk proof)
  • Contact names or email addresses
  • Individual scores and percentiles (private email report only)

Related reading