A city makes WCAG accessibility a mandatory gate before any resident-facing AI UI can be shortlisted. What evaluation practice is this?

Including accessibility requirements as a mandatory gate in civic AI tool evaluation is the correct practice when a city makes WCAG a hard gate before any resident-facing AI UI can be shortlisted. Resident-facing interfaces must be usable with assistive technologies; an inaccessible chatbot effectively excludes residents the city is obligated to serve. Elevating WCAG to a mandatory gate prevents noncompliant products from advancing on the strength of demos that were never tested with screen readers, keyboard-only navigation, or other accessibility criteria. The approach works because accessibility is a civic service requirement and legal risk area, not a post-launch cosmetic fix.

Treating accessibility as a nice-to-have after the marketing launch fails by concept and sequencing: retrofitting accessibility after launch is costly and leaves an exclusionary service live in the meantime. Exempting AI chat widgets from all accessibility standards fails by concept—chat widgets that deliver city services remain resident-facing UI subject to accessibility expectations. Evaluating only keyboard shortcuts for staff and ignoring resident access fails by concept: staff convenience is not a substitute for resident accessibility on public services.

Exam caveat: mandatory WCAG gating means evaluation and shortlist discipline; exact conformance level may follow local policy, but the principle is pass/fail inclusion of accessibility. Keep resident-facing municipal UI framing. Operational check: require VPAT/accessibility evidence and practical assistive-tech testing in the PoC; make WCAG a shortlist gate; block marketing launch of resident AI UIs that fail; and assign remediation ownership before any exception is considered.