Evidence before adoption. Trust before deployment.
SLAT is Novara's criterion-based framework for evaluating sign language AI before an institution buys or deploys it — testing whether a system actually conveys meaning to Deaf users, who is accountable for it, and what happens when it is wrong.
Accessibility-conformance documents and vendor demonstrations certify that a signed output exists — not that it communicates. Institutions are left approving systems they cannot independently verify. SLAT closes that gap with a structured, repeatable assessment built specifically for sign language AI.
The assessment criteria
Evidence, not demonstration
Independent evidence that the system performs as claimed — beyond a controlled vendor demo.
Comprehension, not conformance
Whether signed output actually conveys meaning to Deaf users, not merely that an "alternative" is present.
Governance & Deaf leadership
Whether Deaf expertise holds real decision-making authority in the system's design, evaluation, and oversight.
Accessibility risk & failure modes
What happens when the system is wrong — fallback paths, potential harm, and routes to redress.
Institutional accountability
Clear ownership, monitoring, and responsibility for the system after it is deployed.
Deployment readiness
Whether the tool and the institution are ready for the specific context and stakes of use.
Scope
We define the system, its context of use, and what is at stake for the people it affects.
Assess
We evaluate it against the SLAT criteria using vendor materials, available evidence, and independent review.
Report
You receive a criterion-by-criterion assessment with a clear risk rating.
Recommend
We give a go / conditional / no-go recommendation, with the conditions that would change it.
What your institution receives
- A written SLAT assessment, criterion by criterion.
- A risk rating and a go / conditional / no-go recommendation.
- An independent review of the vendor's claims and evidence.
- Guidance for procurement, contract terms, and ongoing oversight.