AI can now produce code, documents and compliance evidence almost for free. What is scarce is proof that the output is correct. We build software that verifies AI-produced work against the requirements the law sets, starting with Suretto for digital accessibility.
Generation is a commodity. Verification is the product.
01
Doing the work is no longer the hard part.
Language models draft code, vision models read screens, agents run checks. The cost of producing a fix or a report is falling towards zero across every regulated field.
02
Proving it is.
A buyer, an auditor or a court does not want the output; they want evidence that the output meets the requirement. That evidence has to come from something other than the system that did the work.
03
So the verifier never writes the fix.
Every Cano product keeps two systems apart: one generates, one grades, and each result carries the artefact that proves it. People decide the cases that need judgement, and their decisions make the next check better.
The product
A report someone else can check.
This is what Suretto delivers: every claim linked to the method that tested it, the recording that shows it and the version it was tested on. Not a score. Not a badge.
Example
Account settings, billing tab
React app · desktop · 14 Oct 2026
39 checks37 verified2 flagged
All 23 form fields have programmatic labelsVerifiedLabel, name and description read back by two screen readersWCAG 1.3.1
Every control works with the keyboard aloneVerifiedRecorded run: tab, shift-tab, enter, space, arrowsWCAG 2.1.1
Focus order follows the visual orderVerifiedSequence logged for 48 stopsWCAG 2.4.3
Toggle outline is 2.1:1 against the panelFlaggedNeeds 3:1; colour change proposed, awaiting design sign-offWCAG 1.4.11
Illustrative data. A real report carries the recordings, the reviewer’s decisions and the commit hashes.
Products
One product live. The same engine behind what follows.
Each product takes a requirement that companies must meet by law, applies AI to the work, and sells the independent proof.
Pilot programme open
Suretto
Accessibility fixes you can prove.
Suretto repairs accessibility defects in a customer’s own code and verifies every fix with a separate system: automated rules, recorded screen-reader and keyboard runs, and human review of exceptions. The output is an evidence report and, where a buyer needs one, a conformance report (VPAT/ACR).
Procurement teams, universities and public bodies receive conformance reports they cannot test. The Suretto engine turns a vendor’s VPAT into a verified one, line by line, with the evidence attached.
From 11 December 2027 every software and connected product sold in the EU needs verified security documentation. The same two-system engine produces it and proves it.
CrawlAn edge crawler maps the customer’s product as users meet it: templates, components, interactive states, the rendered page and its accessibility tree.
02
DraftLanguage models draft each change inside rules written per defect class. An engineer reviews every draft before it becomes a pull request.
03
VerifyA separate system grades the result: deterministic rules, recorded runs with real assistive technology, and vision checks on the rendered page. It never sees the drafter’s reasoning.
04
ProveEvery result is stored with its method, its artefact and the version it was tested on, so the report can be reissued for any date and checked by anyone.
The company
A product company, built in Nashik.
Cano Solutions Private Limited was established in 2022 and now works on one thing: software that proves AI-made work meets the requirement. Suretto is the first product; the engineers who built it are the company.