What's shipped, what's next
Real, dated engineering changes, not marketing copy. Every entry here reflects an actual change in the product.
Curious what it costs, or want to see it for yourself? Check our pricing or try it.
- Fixed
Fewer false keyboard failures, and counts that match
Custom buttons that work with Enter and Space are no longer reported as keyboard failures. The report header now counts the same findings as the issue list, and anything outside your chosen standards is shown separately.
- Improved
Simpler pricing
One free credit to start, then $10 a credit (₹999 in India), with a small discount on larger packs.
- Improved
VPAT 2.5Rev in the ITI format
VPAT reports follow the ITI VPAT® 2.5Rev template on screen and in the tagged PDF. You can fill in the report details, and criteria that still need manual testing are marked as such.
- Added
127 accessibility laws by country
The standards register now lists 127 laws, each checked against its official text, with the WCAG version it references.
- Added
Compliance profiles
Choose the laws and standards a client has to meet. The report and the PDF then count only the findings that profile covers.
- Fixed
Blocked pages are no longer scored
When a site blocks our scanner or shows a bot check, the scan now fails and the credit is refunded, instead of reporting a score for the block page.
- Improved
Faster pages
Most pages are now pre-rendered and served from the CDN, and large images are WebP, so pages load sooner, especially on phones.
- Added
Scan for free, then unlock the full report
Anyone can scan a page without an account and see a preview of the report. One credit unlocks the full issue list, screenshots and Smart Fix for that scan.
- Added
Real keyboard testing for ARIA Grid, Treegrid, typeahead, and Tree expand/collapse
KITS+ now presses real, browser-trusted arrow keys, Home/End, and Ctrl+Home/Ctrl+End against role="grid" and role="treegrid" widgets, a two-dimensional keyboard model our scanner did not test before. Also added: single-character typeahead search-jump (Listbox/Menu/Tree) and a dedicated Right-Arrow-expands/Left-Arrow-collapses test for Tree View nodes. All eight new checks are registered under WCAG 2.1.1/2.1.3 with an honest confidence tier: one grid-boundary check is flagged medium-confidence, not high, because the WAI-ARIA spec allows two valid boundary behaviors and we can't always tell which one a given grid intends.
- Added
Content-quality detection now understands 11 languages, not just English
Checks like generic link text ("click here"), placeholder headings ("Lorem ipsum"), unhelpful error messages, and generic alt text used to judge every page against English-only phrase lists, so a French link saying "cliquez ici" was missed. These checks now read the page's declared language and check against natural phrase lists in Arabic, Chinese, French, German, Hindi, Italian, Japanese, Korean, Portuguese, Russian, and Spanish, falling back to English only when no language is declared. Built entirely in-house with deterministic phrase matching, with no third-party AI call involved.
- Improved
Standards accuracy pass: added ATAG 2.0, corrected an ADA/WCAG version claim
Added a full Authoring Tool Accessibility Guidelines (ATAG) 2.0 entry, relevant because our own Smart Fix features are an authoring tool under ATAG's definition. Also corrected a page that misstated the DOJ's 2024 ADA Title II rule as requiring WCAG 2.2 (it requires WCAG 2.1 AA) to match our own standards-mapping data, and added a sourced note about 2026 ADA lawsuit-filing trends rather than guessing at a number no one has published yet.
- Fixed
Fixed a payment webhook race that could double-credit an account
Two near-simultaneous Stripe webhook redeliveries for the same event could both pass our old check-then-write logic. Moved to an atomic claim-then-credit pattern (with automatic rollback if crediting fails), matching how our Razorpay and LemonSqueezy webhooks already worked.
- Fixed
Fixed a rare crawl race that could revert a just-computed score back to zero
A crawl’s final summary and score write and its status update were two separate round trips. Under specific timing, the second could read stale data and silently overwrite the correct result. Both writes are now one atomic update.
- Improved
Site-wide accuracy pass: fixed several inconsistent numbers across our own pages
Found and corrected a handful of places where two pages stated the same fact differently (for example our heuristic-group count, our global-law coverage count, and our automated-detection-rate baseline). They are now consistent everywhere, and a couple of overstated claims were reworded to match what we can stand behind.
- Added
Real compliance reports for 13 countries, not just the US/EU
Every jurisdiction listed on /standards now generates a correctly scoped Accessibility Conformance Report, not just an info card. Beyond the existing US (Section 508, ADA Title II) and EU (EN 301 549) editions, you can now generate one for Canada (AODA and the federal Accessible Canada Act), the UK (PSBAR/Equality Act), Australia (DDA), Japan (JIS X 8341-3), Germany (BITV 2.0), France (RGAA 4.1), Italy (Stanca Act), New Zealand (NZWAS 1.1), South Korea (KWCAG), Brazil (eMAG 3.1), Israel (IS 5568), and the UAE, each scoped to its own legally binding WCAG version rather than a relabeled generic report. A new cross-jurisdiction check (Group Z) verifies the published accessibility statement and feedback channel that AODA, ACA, PSBAR, DDA, BITV, RGAA, the Stanca Act, and IS 5568 each require, using real page-scan evidence. Jump straight to any jurisdiction's report from its card on /standards.
- Improved
Crawler now respects robots.txt and rate-limits itself
Site-wide crawls fetch and honor the target site's robots.txt Disallow rules before scanning, and space out page loads across all workers (previously up to 3 simultaneous unthrottled requests). This is safer when scanning a live production site whose infrastructure you don't control.
- Fixed
Stopped leaking internal error details; fixed a silent scan-history failure
Several API routes were including raw internal error messages in their responses to the browser. Those are now only logged server-side. Separately, the scan-history endpoint was turning real failures into an empty "no scans" response instead of reporting them as an error.
- Added
Public API, collaborative remediation, and more
A documented API (POST/GET /api/v1/scans) with webhook support for triggering scans and fetching results programmatically. Shared status, comments and assignment on findings: previously these lived only in each browser's localStorage, so teammates never saw each other's changes. An AI-assisted plain-language rewrite for flagged hard-to-read text, alongside the existing reading-level detection. A standalone, open component-library linter (@accessisight/component-a11y-lint) that catches accessibility issues in React component source code before a page renders, including .stories.tsx files, with an --only-stories flag for a dedicated Storybook CI gate.
- Added
Multi-language platform UI (12 languages)
Navigation, homepage, pricing, settings, and the full sign-in/sign-up page now support English, Hindi, Spanish, French, German, Chinese, Portuguese, Japanese, Arabic, Russian, Korean, and Italian, with a self-built translation system and no third-party translation API. Full-site translation is ongoing: see Roadmap below for what's not covered yet.
- Added
Automated pay-as-you-go checkout
Replaced the manual UPI-QR-code + WhatsApp verification flow with real, automated Stripe Checkout. Added a Single Scan option (₹149) alongside credit packs and the Pro Monthly subscription. Monthly subscription credits now actually renew each billing cycle.
- Fixed
Scanner correctness: dark-mode and dynamic-state checks no longer race other checks
Two heuristic groups that change live page state (color-scheme emulation; real element clicks) were running at the same time as 23 other checks reading that page. They now run one after another, removing a source of intermittent false readings.
- Improved
Faster scans
Dark-pattern and reading-order detection now run in parallel with the rest of Phase 2 instead of after it; linked-PDF analysis runs concurrently instead of one at a time; screenshot-only checks are skipped when screenshots are disabled for a scan.
- Added
Accessibility overlay detection
Scans now flag known third-party "accessibility overlay" widgets (accessiBe, UserWay, AudioEye, EqualWeb, Recite Me) per the accessibility community's Overlay Fact Sheet position.
- Added
Cross-page Consistent Identification (WCAG 3.2.4)
Crawl scans now compare each icon's accessible name across every crawled page, not just within one page, catching a cross-page inconsistency the single-page check couldn't see.
- Added
GIGW (India government accessibility) checks
Seven dedicated checks (bilingual access, Screen Reader Access link, HTML sitemap, text-resize widget, accessibility statement, Help link, content owner and last-updated info) run automatically for .gov.in and .nic.in domains and Hindi-primary-language pages.
- Improved
Real, verified 100% WCAG 2.2 A+AA detection coverage
Reconciled our coverage tracking with what the engine actually detects (it was undercounting real coverage at 75%) and added confidence-tier badges (Automated, Medium confidence, or Proxy signal) so "covered" never reads as "verified with equal certainty." See /standards for the full breakdown.
- Improved
Redesigned VPAT / ACR reports
Reports are now grouped by WCAG principle instead of one 55-row table, with search, edition-aware standards tables, and each row's detection-method caveat carried through into the remarks, not just shown on /standards.
- Added
Defect-to-VPAT (bring your own findings)
Already have manual audit findings? Import them directly and generate a structured VPAT, no scan required.
Roadmap
What we know is still missing, roughly ordered by what we're tackling next. Effort is a rough size, not a promised date.
Full-site translation
Extend the 12-language translation system (already live on navigation, homepage, pricing, settings, and sign-in) to the dashboard, scan results, standards library, and VPAT report pages, which still show English regardless of the selected language.