Language: Français English
Lab program — applications open

An R&D lab for digital accessibility. Your site could be audited for free.

Accessibility Sentinel builds automated detection methods for WCAG and RGAA compliance — automated scanners, real evidence (screen reader, session timeout, vision triage), independent evaluation of 162 criteria. To refine these methods on real cases, we're opening a limited number of complete, free audits: 100% of the features, in exchange for an anonymized case study and feedback.

How the program works

Your site is the fuel for our lab

A research lab needs real cases to validate its methods — not mockups. Rather than billing for our first audits, we give them away in full, in exchange for use as a case study and honest feedback.

100% of the features

The same audit we'd otherwise charge for — no watered-down version: automated scanning, real evidence, 162 WCAG + RGAA criteria, full deliverables.

Two simple commitments

Authorize an anonymized case study and give us feedback — no payment requested.

A limited number of spots

Every application is reviewed; we favor real sites and a diversity of sectors and tech stacks.

Apply for a free audit See the selection criteria

On our reference project

An independent audit of a high-traffic transactional website — the figures below come from that real project, detailed in our case study.

81% RGAA · 71% WCAG Share of criteria covered by our automated approach. On RGAA: 20% with no human check at all, 61% with a quick manual confirmation (the analysis is already done — validating or refuting an established verdict, never a re-audit). The rest (19% RGAA, 29% WCAG) falls outside what we automate today: as a research lab, we aim to push the limits of automation as far as possible — combining traditional methods with LLM judgment for human-style analysis — rather than falling back on a classic manual audit. Full breakdown on the Capabilities page.
70 Documented defects (46 on the main journey + 24 from a cascading audit of 174 pages), each with severity, WCAG and RGAA criterion, annotated screenshot.
162 Regulatory criteria evaluated independently (56 WCAG A/AA + 106 RGAA 4.1.2), each with a verdict and test protocol.

What sets us apart from standard scanners

Accessibility Sentinel grew out of an R&D lab building automated detection methods for digital compliance — we're not a certification body, we surface concrete evidence and testable protocols your team can act on. Most audit tools stop where automation stops: they cover a fraction of WCAG/RGAA criteria, and the rest gets tagged "manual check required" and left there. We start from the opposite premise.

Real evidence instead of assumption

The speech a real screen reader (NVDA) actually says, an actual 30-minute wait to test a session timeout, a real simulated touch gesture — not heuristics approximating user behavior.

Every "manual" criterion stays actionable

For every criterion that can't be automated, a precise test protocol (tool, actions, pass/fail criterion) instead of a plain tag — a human auditor can pick up the work without starting from scratch.

WCAG and RGAA, two standalone frameworks

56 WCAG 2.1/2.2 A/AA criteria and 106 RGAA 4.1.2 criteria tested each on its own terms — a simple mapping table produces both false positives and false negatives. The RGAA is due to evolve into a version 5, currently being drafted.

Honest about the limits

Screen-reader parity, a payment step never gone through, alternate-version fidelity: we document what remains fundamentally manual instead of hiding it.

Every verdict carries its evidence AND its method's limit

Our multimodal vision/language model's judgments are counted separately (dual rate displayed: official vs. mechanically proven) and verifiable by sampling, rather than diluted into a single score.

Capability catalog

Large-scale automated detection, real evidence, scaling to a whole site, and deliverables ready for your teams.

Automated detection

axe-core 4.9.1 scanning in a real browser, DOM inspection (accessible names, heading structure, ARIA roles), automated keyboard navigation, contrast, 200% zoom, text spacing, office documents.

Real evidence

A bridge to NVDA with exact quotes of the speech heard, a real session-timeout test, vision-model-assisted triage for unlabeled icons, touch-gesture simulation.

Scaling to a whole site

Parallel automated crawl to audit hundreds of pages in cascade, aggregating systemic defects (shared component rather than isolated symptom), explicit scoping.

See the full catalog →

Deliverables ready to use, in French and English

Structured CSV & Excel workbook

One finding per row, normalized columns (severity, WCAG criterion, RGAA criterion, detection method). Multi-tab workbook with native formulas, not hardcoded values.

Word report & interactive chat

Executive summary, detailed findings, annotated screenshots at exact DOM coordinates. Chat interface to query results in natural language, citing finding IDs.

Every deliverable ships as two separate files — French and English — not a single mixed-language document.

Ready to know where your site stands?

Tell us about your site and any regulatory deadlines (EAA, RGAA...) — we'll come back with a clear scope.