Platform

Digital accessibility operations, not just a scanner

Verassa is the evidence system for finding, deciding, fixing, verifying, reporting, monitoring, and improving accessibility work across digital surfaces. The built web core is live first; broader lanes stay clearly labelled until their gates clear.

Book a demo Read the methodology

Evidence engine

Sample

Diagnostic pass

A barrier is detected

CriticalWCAG 2.1.1

Checkout step cannot be completed with a keyboard.

The diagnostic surfaces a candidate. On its own that is just a rule trip — not yet something a team can act on or defend.

axe + interaction replay
› target  #checkout-next
› event   keyboard-trap
› impact  serious

Surface status

Whole-app lanes, one trust model

The product story names the full operating system while keeping every lane honest about its PRD §11 surface trust state and what has to clear before customer exposure.

Trust boundary

One product map, explicit exposure controls.

Verassa can show the direction of the whole operating system without turning planned, candidate, or internal work into a claim. Every lane carries its surface trust state and the gate that keeps customer exposure honest.

No lane is labelled Validated or External-reliance ready until the relevant PRD §11 gate and human-review controls support that language.

Built: Web scans, review, evidence, reports
Held: Remediation, ACR, Defense Pack exposure
Named: Developer, mobile, document, media expansion

Surface status map

PRD v5.2 trust states by lane

Planned
Built
Candidate
Validated
External-reliance ready

Candidate
Web & authenticated-flow audits
Scoped browser runs, authenticated-flow foundations, evidence capture, findings, review, reports, and exports.
Exposure control
PRD §11.2 real-site diagnostic floor
Outputs are shown conservatively and routed to review.
Built
Review & Decision Layer
Qualified reviewers confirm, dismiss, defer, route, and attest decisions with rationale and audit history.
Exposure control
Reviewer authority retained
Code can run internally or in reviewer-assisted workflows.
Built
Remediation & verification
Source-aware fix planning, verification re-scans, rollback context, and customer exposure controls.
Exposure control
Release-quality clearance + verification records
Code can run internally or in reviewer-assisted workflows.
Planned
Developer Hub / MCP
CI checks, pull-request evidence, issue routing, and developer tools tied to verifier-first workflows.
Exposure control
Verifier-first workflow evidence
Product lane is committed, but not yet supported.
Planned
Continuous monitoring
Recurring scans, regression alerts, ownership routing, and outcomes signals for changing products.
Exposure control
Recurring-scan evidence and alert policy
Product lane is committed, but not yet supported.
Planned
Native mobile, document, and media studios
Mobile apps, PDFs, Office files, captions, transcripts, audio description, and media players get their own evidence models before launch claims.
Exposure control
Surface-specific evidence models
Product lane is committed, but not yet supported.
Planned
Procurement & ACR Center
Questionnaire support, ACR drafts, stale-evidence warnings, and external-reliance outputs require evidence and human sign-off.
Exposure control
Real remediation evidence + reviewer sign-off
Product lane is committed, but not yet supported.
Planned
Enablement & maturity
Training assignments, maturity scoring, advisory workflows, and repeat-defect insight tied to actual findings.
Exposure control
Outcome history and advisory review
Product lane is committed, but not yet supported.

Operating facts

Evidence first, status always visible

Platform-configuration facts — not customer-outcome claims.

Web core: Built; Evidence, review, findings, reports, and authenticated-flow foundations.
Expansion lanes: Labelled; Planned and gated areas are named that way until their own evidence gates clear.
Trust record: 1; Evidence, reviewer rationale, verification state, and provenance travel together.

Why an operating system

A scanner reports. An operating system moves work.

An automated scanner ends at a list of issues. Most accessibility work begins there: scoping assets, deciding which findings are real, assigning ownership, planning fixes, verifying outcomes, and keeping evidence fresh.

The platform is built around that work. A finding has a state, an owner, evidence, a rationale when it is dismissed, and a verification record when it is fixed. The result is not a longer list — it is a shorter distance between a barrier and its removal.

The agents

Three agents and a human-review layer

The Diagnostic Agent evaluates an approved scope and captures evidence for each candidate finding. The Decision Layer is where a qualified reviewer makes the judgment calls a model should not make alone. The Remediation Agent remains gated from customer-facing claims until the diagnostic and verification gates clear.

Agents do the production work: the repetitive, high-volume evaluation, routing, drafting, and verification support. People keep the judgment. The platform is explicit about which is which, on every finding and every lane.

See how it works, step by step

See the operating system around a real accessibility workflow

A demo walks through the current web core and shows how evidence, review, decisions, remediation planning, verification, and reports stay connected.

Book a demo Read the methodology