Lythe

Authorization & limits

What you may test, and how to read results responsibly.

Two things determine whether a Fraser run is legitimate: whether you were allowed to test the target, and whether you represent the output honestly.

What you may test#

You may use Fraser only on websites, applications, APK files, product flows, and digital assets that you own, operate, control, or are expressly authorized to test.

You must not use Fraser to test, scrape, monitor, access, or analyze third-party websites, applications, systems, or content without proper authorization.

Before pointing Fraser at something, confirm your use doesn't violate applicable law, third-party terms of service, contractual restrictions, intellectual property rights, privacy rights, or security requirements.

Data you upload#

You are responsible for ensuring that material you submit, URLs, screenshots, APK files, call recordings, transcripts, CRM records, customer feedback, was collected lawfully and may be shared with Lythe for this purpose. If you're generating personas from customer conversations, that includes having the rights and consents to do so.

Lythe does not permit third-party AI or model providers to use your data to train their models. See the Privacy Policy and Subprocessor List.

How to represent output#

Simulation outputs, persona responses, visual evaluations, and bug-testing results are generated for product evaluation and research support.

Do not present them as:

  • real user interactions or customer interviews
  • verified vulnerability reports
  • statistically valid survey results
  • guaranteed market truth
  • a legal accessibility conformance audit or certification (for example WCAG, ADA, or EN 301 549)

unless you have independently validated them through appropriate methods.

Where synthetic evidence runs out#

Fraser is strong evidence that something is confusing, and weak evidence about how much revenue it costs you. It models behavior from patterns; it doesn't observe your market.

When the question is "is this flow broken?", Fraser answers it well. When the question is "will this segment pay for this?", Fraser can generate a hypothesis, and you should go test that hypothesis with real people.

Operational limits#

Fraser is for evaluation, not load. In particular:

  • Not for load, stress, or availability testing. Do not use runs to generate load against a target or to measure how much traffic it can absorb.
  • Request rate is capped per target. Fraser limits how fast and how concurrently it drives any single target.
  • No transacting or account creation on systems you do not control. Do not use runs to place orders, submit real payments, or create accounts on anything you do not own or operate.
  • Proof of authorization may be required. Lythe may ask you to demonstrate that you own or are authorized to test a target before or during a run.

Related#

Was this page helpful?