How WebLens works
WebLens runs an assessment in five stages. You bring an application and a user journey; we return a structured, evidence-backed report.
The five stages
- Add your application. In WebLens, click "Add application" and paste your app's URL, public or behind a login, we handle access on the way in. This just registers the app; no assessment runs until you start one.
- Define your user journeys. Sign-up, checkout, a configurator, the paths your users actually take. See Writing user journeys.
- We assess. AI agents walk every journey, capture evidence at each step, score against the criteria set, and produce a recommendation for every issue surfaced.
- You check the report. A grouped, evidence-backed report. Each finding links to a criterion, a confidence tag, and a suggested remediation.
- Track your quality journey. Re-run assessments over time and watch your UI/UX Health Score climb release after release.
What you provide
- A starting URL (or a list of URLs for multi-flow runs).
- A user journey, the steps a user takes through the application.
- Optionally, a list of criteria to skip (e.g. if you've already run a manual audit).
What you get back
A report grouped by category (UI, UX, Accessibility) and subcategory. Each finding includes:
- The criterion that flagged it.
- Its confidence tag, so you know how much to trust the AI's call.
- A screenshot or trace as evidence.
- A suggested remediation, when the criterion supports one.
How each criterion is assessed
Every criterion runs over the same evidence: the agent walks the user journeys you define, capturing a screenshot and the page state (DOM, attributes, computed styles) at each step. Criteria then reason over that captured evidence, what changes per criterion is the question being asked, not the navigation.
Most criteria, including all Usability criteria and the majority of Accessibility criteria, are evaluated with AI-powered vision analysis: the agent inspects the captured screenshots and page state and judges the criterion against a written rubric. Some Accessibility criteria additionally use deterministic measurements where the property is objectively computable, for example contrast ratios of text and UI controls, or whether a page declares a language. This is reasoning over real rendered evidence, not a static rules engine run blindly against markup, which is what lets WebLens cover usability and cognitive criteria that automated rule libraries cannot.
For accessibility, each WCAG success criterion is broken into sub-tests, and WebLens is explicit about which it covers fully, which only partially, and which still need a human. A few requirements, CAPTCHA alternatives, caption accuracy, and parts that demand subjective judgment, are intentionally left to manual review. The sub-test breakdown on each criterion page documents these boundaries, and we validate every criterion against expert review rather than asserting accuracy, see Accuracy & confidence.
Your report: when it's ready, sharing, and reviewing
- When it's ready. A report is generated automatically as soon as an assessment finishes. You'll find it under the application's Reports tab; while the assessment is still running, it shows as in progress.
- Sharing. Open any report in the browser and use Copy link to share it with stakeholders. They open the live report directly, no PDF or export needed.
- Reviewing. Issues are grouped by user journey and by criteria. As you work through them you can mark issues as read, so the team can see at a glance what's already been triaged.
You stay in control of the results
WebLens does the heavy lifting, but the findings aren't locked. You can review, edit, dismiss, or confirm any issue it surfaces, so the final report reflects your judgment, not just the agent's. Most teams rely on the high-confidence findings as-is and confirm the rest.
Next
Read about the criteria we evaluate, or see how we set accuracy & confidence.