The agent experience platform

AI agents are reading the web
for your customers now.
Can they read you?

Every recommendation an AI assistant makes starts with a machine reading your site — no JavaScript, no images, raw HTML only. We run the deterministic engine that measures exactly what those machines see, proves what they miss, and watches it every single day. Any website. Evidence for every claim.

free · no signup · under a minute · deterministic · evidence included
your-store.com · agent purchase walk DEMO
Alpine Merino Sock
$19,90
SML
Add to cart
OKFound the product — “Alpine Merino Sock”
OKPrice readable — “$19,90”
OKVariants distinguishable — S / M / L
!!Shipping terms — nothing I can quote
VERDICT: abandoned at “shipping terms”.
Run this walk on YOUR store →
OK machine-readable · !! blocks the purchase demo walk · the real simulation quotes YOUR store's pages
Agentic commerce

Is your site invisible
to AI agents?

ChatGPT, Perplexity & Co. read stores without JavaScript, parse product schema and check policies before they recommend anything. A store they can't parse doesn't exist for them — and most owners have no idea.

Fix it — the Agent-Ready Kit, $19 →
FREE CHECK ↑ · DETERMINISTIC CHECKS · EVIDENCE INCLUDED · SAMPLE REPORT
agent view · sample-store.example DEMO
> robots.txt for GPTBot, ClaudeBot, PerplexityBot… ALLOWED
> product JSON-LD (price · currency · availability)… MISSING
> prices without JavaScript… VISIBLE
> return & shipping policies… LINKED
> llms.txt… NOT FOUND
> title, meta & Open Graph… OK
READINESS 49/100Invisible to AI shoppers
sample store · your check runs live in under a minute

What you get
at every step.

See what breaks, get the code that fixes it, then keep it fixed. Each step builds on the last — and every line is computed from your own pages, never invented.

STEP 1 · THE FREE CHECK

Both views. Three verdicts. Evidence attached.

The engine fetches your site twice — raw, exactly as an agent reads it, and rendered, as a browser paints it — and classifies every signal into one of three states: present, missing, or the one nobody else measures: present but invisible — on your page, but only after JavaScript runs, so a model never sees it. Blocked crawlers gate the score entirely, with the exact robots.txt lines to delete.

  • Dual-view differential analysis — raw vs rendered, per check
  • Truth checks: does your machine data match your visible page?
  • Page-type aware scoring — store, catalog or SaaS, each on its own 100
Run the free check →
dual-view analysis · your-store.com SAMPLE
STATE · CHECK · VIEW EVIDENCE
PRESENT  Product schema · valid in raw HTML
JS-ONLY  Prices · rendered view only — agents never see them
MISSING  Own agent instructions · platform boilerplate served
PRESENT  Policies · linked in raw footer
sample output · your check computes live from your pages
STEP 2 · THE AGENT-READY KIT · $19

The fix, generated from your own pages.

Not advice — files. Paste-ready product schema for every page we found, built from your own observed values with anything unknowable marked instead of guessed. Your own agents.md and llms.txt, replacing the boilerplate your platform serves to every site. A numbered fix order. One download.

  • One schema block per product page, values read off your pages
  • agents.md + llms.txt as real files, yours not the platform's
  • The agent purchase walk: the exact step a machine gives up
Run the Kit →
frictionless-kit.zip · contents SAMPLE
FIX-ORDER.md
agents.md
llms.txt
schema/01-alpine-merino-sock.json
schema/02-trail-runner-low.json
schema/… one per page we found
every value read off your pages · unknowns marked REPLACE-*, never guessed
STEP 3 · FRICTIONLESS WATCH · $49/MO

Then we watch it. Every day. With proof.

A theme update, a replatform, one WAF rule — the machine layer breaks silently while the page looks perfect to humans. Watch re-runs the full engine on your domain daily and alerts you the moment a check regresses, with the evidence in the alert. And a fix only counts when the next run confirms it — measured, never self-reported.

  • Daily full-engine runs, both views, non-flapping alerts
  • Webhook alerts that name the check, both states and the delta
  • Every run feeds the public agent-readability index
Start watching →
watch alert · webhook SAMPLE
your-store.com: 1 check regressed (score 82 → 41)
!! Product schema (JSON-LD): present → missing
theme deploy detected at 14:32 · raw HTML no longer carries the block
ok next run confirms your fix — or keeps alerting
sample alert · real alerts quote your own checks
Simple pricing

Measure free. Fix once.
Then watch it, daily.

One deterministic engine under everything: the free check measures, the Kit hands you the fix as files, and Watch re-runs the whole engine on your domain every day — with an alert the moment something breaks and the fix proven by the next run.

Enterprise AI-visibility platforms start around $500/month. Frictionless is deterministic to the core — same input, same score, no model in the loop — which is why it can be this fast, this checkable, and this priced. The $19 Kit carries a 30-day money-back guarantee.

The Check
Free
no signup · under a minute · any website

The full engine, once: both views, three states per check, evidence for every claim. Forever free.

  • Dual-view analysis: raw HTML vs rendered — per check
  • Truth checks + the ACP/UCP protocol rails
  • Crawler gate with the exact robots.txt lines to fix
  • A preview of your generated product schema — your real data
  • One competitor comparison, side by side
  • Your living Agent Report — a permanent link that re-checks on every view
Run the free check →

GDPR-compliant · No signup · No credit card

Agent-Ready Kit
$19
one-time payment · 30-day refund

The fix as files, generated from your own pages. Paste it in and you are done.

  • Paste-ready Product JSON-LD, one block per page we found
  • Your own agents.md + llms.txt — not the platform boilerplate
  • Agent-ready title, meta and FAQ copy per page
  • The purchase walk: the exact step an AI buyer gives up
  • Numbered fix order + everything as one ZIP
  • Re-check included: run it again and watch the score move
Get the Kit →

If it shows you nothing new, you get your money back.

★ The platform
Frictionless Watch
$49/month
daily monitoring · any website · cancel anytime

The machine layer breaks silently — a theme update, one WAF rule — while the page looks perfect to humans. Watch runs the full engine daily and tells you the moment it happens.

  • Daily full-engine runs — both views, all checks
  • Non-flapping alerts that name the check, both states, the delta
  • A fix only counts when the next run confirms it
  • Score history + event log on a permanent dashboard
  • Webhook alerts (Slack-compatible), no account needed
  • Every run feeds the public agent-readability index

Straight to Stripe — no account, no email. The link you get back is your dashboard.

Run a Shopify store? The Frictionless app adds the loop inside your admin — weekly re-checks with score delta, the Weekly AI-agent readiness re-check, one-click theme blocks. $29/month with 7 days free, or $290/year, billed by Shopify. Never writes to your theme or products.

Ready to see what
AI agents see on your site?

Under a minute · evidence included · no signup

Check my AI visibility →
or explore the full live demo →
FAQ

Common questions

AI assistants read websites the way machines do: raw HTML, no JavaScript, no images. What they can read decides what they recommend. Frictionless measures exactly that experience with a deterministic engine, hands you the fix as files, and watches your domain daily so a breakage never goes unnoticed. Any website — stores, SaaS, courses, publishers.
Three things. It fetches your site twice — raw and rendered — and reports three states per check, including the one nobody else measures: present on your page but invisible to machines. It runs truth checks: whether the data your site serves machines matches what humans see. And it is deterministic — same input, same score, no model in the loop — so every claim is checkable and every fix is provable by the next run.
Yes. Full engine, both views, evidence included, no signup, no credit card. The permanent Agent Report link re-computes on every view — bookmark it, nothing to unsubscribe from.
The machine layer breaks silently: a theme update, a replatform, one WAF rule — and your schema, prices or crawler access are gone while the page looks perfect to humans. Watch re-runs the full engine on your domain every day and alerts you the moment a check regresses, naming the check, both states and the score delta. A fix only counts when the next run confirms it. $49/month, cancel anytime, no account — the dashboard link is the access.
The fix as files, generated from your own pages: paste-ready Product JSON-LD for every page we found, your own agents.md and llms.txt instead of the boilerplate your platform serves, agent-ready copy per page, the purchase walk showing where a machine gives up, and a numbered fix order — all as one download. 30-day money-back guarantee.
Any website. The engine classifies the page first — store with products, store without a reachable catalog, or a non-store like a SaaS or course site — and scores each on its own 100 points, only from checks that apply to it. A SaaS is never penalised for having no product schema. Shopify stores additionally get the $29/month app with the loop inside the admin.
We read publicly served pages, exactly as any crawler would — respecting robots.txt. Scans store what your site serves to machines, per check, so your history and the public aggregate index exist. No customer accounts, no personal data collection on the free check, no email required anywhere.