Agent Journey Lab

Research

Our research boundary

We study whether AI agents can finish public customer journeys. This page states where a public run stops and how an authorized test is different.

For public research, we stop before we enter data in a form. Authorized tests use the written scope that you approve.

Public research

Public research uses pages that any visitor can open. We take no consequential action.

  • We read public pagesThe agent opens pages, reads content, and follows public links.
  • We stop at the formThe run stops before form entry, account use, submission, payment, booking, or a claim.
  • We declare the agentOur sessions declare AgentJourneyLab/0.1 in the user agent string.
  • We keep the rate lowWe use a low request rate and we respect robots.txt.
  • We keep results privateWe do not publish a result about a named site without written permission.

To ask a question about a public run, use this address. To ask us to exclude your domain from public research, send the domain in that email.

research@agentjourneylab.com

The difference between public research and an authorized test

  • Public research

    • No agreement and no payment.
    • Public pages only.
    • The run stops before any form entry.
    • The work supports our own study of failure patterns.
  • Authorized customer test

    • A written scope that you approve, and payment before the test.
    • You confirm your authority for the named domain.
    • The run can go deeper, but only inside the limits that you set.
    • The record belongs to you and stays private.

Read the pilot scope