Give us a job to do.
One public URL. Up to three things a visitor should be able to accomplish. Submit from your workspace, API, or MCP.
Find the confusing bits before your customers do. Send a website and three tasks. Get honest, human feedback your agent can act on.
Now collecting early-access signups. Paid tests are not open yet.
“Can someone actually use this?”
“I knew what you were selling.
I didn’t know where to click.”
Join the free list for our first human usability tests. We’re gauging interest before opening the paid pilot.
Planned pilot price: $5 per test. Joining the list is free, with no obligation to buy.
Your tests can pass.
Your users can still get stuck.
A small, practical reality check.
No research project required.
One public URL. Up to three things a visitor should be able to accomplish. Submit from your workspace, API, or MCP.
A real person spends up to 10 minutes working through the tasks on desktop or mobile and notes where they struggle.
Task outcomes, specific observations, and prioritized fixes. Read the report in your workspace or retrieve it as JSON.
A useful report gives your agent something concrete to fix. Every finding includes what happened, how to reproduce it, and a suggested improvement.
See the report formatThe offer was clear. Getting started wasn’t.
On mobile, I scrolled past the text link twice before realizing it was the next step.
You’re working directly with a human tester. We’re starting small to learn what makes feedback useful to agents and the people building with them.
Free to join. We’ll contact you when the pilot opens.
Check availability. Submit tasks. Retrieve the report.
The feedback is human. The handoff is structured.
request_test({
url: "https://your-site.com",
tasks: [
"Find the price",
"Start the signup flow"
],
device: "mobile",
idempotency_key: "launch-review-001"
})
// Human test → structured reportWe’re collecting interest first. Join early access for free; paid testing is not open yet. Signing up does not book a test or authorize a charge.
Yes. A real person performs the pilot tests and writes the feedback. The report is not an AI-generated review.
A public website you own or have permission to test. No account credentials, confidential data, real purchases, or destructive tasks. This is usability feedback, not a security audit.
No. This first pilot delivers written observations and a structured report. Keeping the format small lets us learn what is actually useful.
Tests are performed manually. The pilot accepts only three active orders at a time. Track your request in the workspace; there is no guaranteed delivery deadline during this early pilot.
The first version returns a Stripe Checkout link. An authorized payer completes checkout before testing starts. Autonomous wallet payments are not supported in this pilot.