Skip to main content
Design-partner cohort open · evidence-backed accessibility operationsApply
Verassaerassa
MethodologyPricing
Sign inBook a demo
Loading…
Verassaerassa

Evidence-backed digital accessibility operations with qualified human review, careful claims, and exportable proof.

Private beta

Apply to evaluate the platform →

Platform

  • Overview
  • How it works
  • Methodology
  • Evidence packages
  • Pricing

Workflows

  • Product and engineering
  • Legal and compliance
  • Accessibility consultancies
  • Ecommerce

Trust

  • Trust Center
  • Security
  • Privacy
  • Claim boundaries

Learn

  • Docs
  • Research
  • Changelog
  • State of Web Accessibility

Company

  • About
  • Contact sales

AI-augmented accessibility evaluation. Findings accelerate audit production but do not replace qualified-reviewer attestation. No scan is a conformance guarantee.

  • Policies
  • Accessibility statement
  • Refused use cases

© 2026 Verassa. Evidence-backed accessibility operations.

Product

Evidence Packages

A finding you cannot inspect is a finding you cannot trust. Each one is captured with the evidence a reviewer needs to confirm it.

Book a demoWhy scanners are not enough →

Verassa evidence protocol

  1. Evidence

    01

    Screenshot, DOM, replay, and axe baseline captured before decisions.

  2. Judgment

    02

    Reviewer route, rationale, and owner stay attached to lower-confidence work.

  3. Verification

    03

    Re-scan records and disclaimers travel with reportable outputs.

What is captured

What is in an evidence package

  • 01

    Screenshot

    The barrier in context, annotated where it helps.

  • 02

    DOM snapshot

    The exact markup that was evaluated.

  • 03

    Accessibility tree

    What assistive technology actually receives for the element.

  • 04

    Interaction replay

    The keyboard or pointer steps that reproduce the issue.

  • 05

    Transcript

    The agent's reasoning behind the finding, recorded.

  • 06

    Automated baseline

    The axe-core result, retained as a baseline layer beneath the agent findings.

By the numbers

Captured carefully, by construction

Platform-configuration facts — not customer-outcome claims.

Evidence types per finding
6
Screenshot, DOM snapshot, accessibility tree, replay, transcript, baseline.
Default-on redaction pass
1
PII is stripped before evidence is displayed or exported. Not skippable.
Findings without provenance
0
Each one names the agent, model, and ruleset versions behind it.

Redaction

Captured carefully, redacted by default

Evidence is the page, and a page can carry personal information. The evidence pipeline redacts personal information before evidence is displayed or exported. Redaction is the default and is not skippable; a sensitive-scan mode increases it.

Where a finding's evidence was redacted, the report says so, so a reviewer understands the chain-of-custody implication rather than discovering it later.

Read about privacy →

Inspect the evidence behind a finding

A demo opens an evidence package and walks through what the agents captured.

Book a demoRead the methodology