Features
The machinery around the shoppers: guardrails on money and access, hard cost ceilings, sharing, trend lines, and an MCP server so your AI assistant can run the whole loop. Everything here ships on the experience-feedback product you already saw — how it works →
Aim a run at one URL — the checkout you just shipped, a new signup flow — and shoppers walk exactly that.
You confirm you own or are authorized to test each site you add — nothing installed, no DNS record, no snippet.
Agents stop at payment forms by design. Real checkouts only on staging, with test cards you provide.
Every run shows its cost estimate before launch and enforces a budget ceiling, step caps, and timeouts.
Shoppers hit unlabeled buttons and keyboard traps the way real visitors do — and every page they visit gets a formal WCAG A/AA scan, exact violations listed worst-first in the report and PDF.
Every run can mint a public replay link — anyone watches the shoppers' journeys, screenshots plus reasoning, no account needed. Revocable in one click.
Health-score trend across your last 30 runs, plus which competitors the AI assistants keep naming instead of you.
A built-in MCP server: Claude, ChatGPT, and Cursor can add sites, launch shopper tests, and read reports — your assistant literally hires our shoppers. On Claude it's one OAuth sign-in, no key to paste.
Pick a before and an after — any two finished runs — and see both scores side by side with every finding sorted into resolved, still there, and new.
On Scale, client-facing PDF reports carry your brand name and logo instead of ours — built for agencies whose clients should see the agency on the deliverable.
All of it is on the free tier — see it on your own site
10human shoppers free every month tell you why they'd leave. No card, no trial clock.