Better UX. More conversions. After every release.
Tell Corexi what a win is for your product and your market: sign-ups, activation, upgrades, checkouts. It signs in, walks your app like a new user, and shows what stands between people and that goal, with the evidence, what it is costing you and the fix.
One engine. One loop, every release.
The engine reads what your product looks like and what people actually do in it, together, not in two tools. It prices what it finds, hands over the fix and checks it on the next run. The stops below are this loop, in the order a run takes it.
- 01 · Sense+ your behavioural data
- 02 · FuseVisual78Behavioural69
- 03 · DiagnoseHIGHSend invoice reads as disabled+2.4%recoverable / mo
- 04 · Fix & shipLED-214#482MCPUX Score 69 → 72 next run
↻ run #18 after your next release
What only Corexi does
Our methodology →
The Neurodiversity Lens scores each screen as people with dyslexia, ADHD, autism, sensory sensitivity or a colour-vision difference meet it. Simulated on the screen, not described in a persona.
Heatmap tools show what happened; audits show what a screen looks like. The engine reads both together, so a finding says what is wrong, how many people it stops and why.
It walks again after every release and checks itself against the last run. New rules and lenses in the engine reach every product without a new project.
Tell it what a win is. It tailors everything to it.
Pick the goal (conversion, activation, retention or revenue) and the market you are in. Autopilot walks the steps that matter for it first, and ranks every finding by it. Goals and benchmarks →
- One goal, everything weighed by it
- An activation checklist
- Your sector and your rivals
- 1Sign up100%
- 2Verify email91%
- 3Add first client64%Biggest drop. Autopilot walks this step first and ranks what blocks it.
- 4Create first invoice41%
- 5Send first invoice23%
50avg
65top
82You 72
Billforge leads on onboarding: a sample invoice is ready at sign-up.
It signs in and walks your product like a new user.
Menus, lists, forms, reports, settings. Each flow to its end, every step written down, so you can see how it got to everything it found. How Autopilot walks →
- It gets in
- It never clicks what it shouldn't
- Desktop or phone, per login
- 001Sign in/login
- 003Dashboard/dashboard
- 005Invoices/invoices
- 009New invoice/invoices/new
- 011Errors/invoices/new
- 013Add client/invoices/new
- 018Clients/clients
- 024Billing/billing/plan
- 029Team/settings/team
- 025⊘ Upgrade & paynever clicked
- 030⊘ Invite peoplenever clicked
Every screen it reaches, read closely and marked up.
The screens behind the login, at each width, scored one by one, plus the dialogs, focus and error states that only exist after someone clicks. What it checks →
- Screens and their states
- WCAG 2.2 AA, measured
- Evidence you can check
Visual and behavioural, fused into one diagnosis.
Corexi reads what the screen looks like to five kinds of mind and what real people did on it, fuses the two, and tells you why it happens, what it costs and how to fix it. How the score works →
In plain wordsFuse means one verdict from how a screen looks and what people actually do on it. The Neurodiversity Lens reads the screen as people with ADHD, dyslexia, autism, sensory sensitivity or a colour-vision difference meet it.
- Five lenses, each scored
- One diagnosis from both halves
- Priced, with the fix written
The fix lands where your team already works.
Your editor fixes it in place, your tracker gets the ticket with the evidence, and your channel hears about a regression first. Corexi over MCP →
In plain wordsMCP is how coding agents such as Claude Code or Cursor talk to tools. Corexi gives them ten: to check a design or a diff before it ships, and to follow a release through.
- In your coding agent
- Jira and GitHub
- Slack, Teams and email
After every release, it walks again and tells you what moved.
Each run compares itself to the last: the score, the screens, the findings fixed, new or back. Your UX debt goes down where everyone can see it. Runs and regressions →
In plain wordsUX debt is every open issue, weighted by how much it hurts, added into one number, so you can see it shrink.
- After every deploy
- Regressions first
- UX debt you can pay down
- v2.1370 · steadyMon 14:02
Two new low findings on Reports. Nothing dropped.
- v2.1461 · dropped 9Wed 11:40
Plan prices vanish on mobile. Alert in #product within the run.
#product · alert - v2.1568 · fix shippedThu 16:15
Pull request #482 from the finding; merged the same afternoon.
ledgerly/web #482 - v2.1671 · verifiedFri 10:05
The next runs agree: the finding closed itself.
- v2.1772 · +1Mon 09:12
5 fixed, 2 new, 1 came back, listed screen by screen.
Before it ships, and after it's live.
One engine on both sides of a release, so what your developers check before a merge is what the product is measured on after it.
Your coding agent asks Corexi first.
Check a screen, a component or a diff against the same rules and lenses, from Claude Code, Cursor, VS Code or Windsurf over MCP. Issues get fixed before they merge.
Autopilot walks the live product, every release.
One call from your pipeline, or a schedule. It signs in, reads the product with your analytics, and tells you what moved: fixed, new and came back.
Behind login, 2FA and roles.
A login set to phone width walks it at 390px.
iOS and Android apps on web views, walked.
Scored from the screens you upload.
The European Accessibility Act applies to your product. Probably.
Digital products sold to people in the EU must now be accessible, and each country sets its own checks and fines. Corexi keeps the evidence for you: every screen, behind the login too, measured against WCAG 2.2 AA and checked again after every release.
- Do you sell to consumers in the EU?Anywhere in the 27 member states, from anywhere.
- Is your product one of these?Online shop or marketplace, banking or payments, e-books, passenger transport tickets, phone or messaging services.
- Do you have 10 or more staff, or over €2M a year?Smaller companies offering services are exempt.
A simplified reading of the Act, not legal advice.
Teams that ship continuously.
Product, growth and engineering teams running Corexi after every release.
eWorks 2026 · Acceleration Program
From quarterly audits to daily Autopilot runs. Fix-ready code shipped straight into their sprint cycle.
Continuous UX guardrails across 4 connected modules. One UX Score replaced module-by-module design QA.
UX quality layer for production-ready apps. Monitored across every deployed product.
“Corexi found 14 UX issues on our checkout in 10 minutes. Things we hadn't noticed in months of user testing. The fix-ready code saved our team days of work.”
“Corexi gives us a real-time UX pulse across every buyer and seller flow. We ship faster knowing friction is caught before it costs conversions.”
Safe to let loose on your product.
Stored encrypted and used only to sign in. We recommend a test account per role.
Save, pay, delete, invite, sign out, clock in and security settings, plus any words you add.
Every step is live on screen. Take the browser over at any moment, or stop the run.
It looks and clicks like a new user. It does not change your data, your code or your settings.
Is it safe to give Corexi a login?
Yes, and a test account is best. The login is encrypted, used only to sign in, and the run never saves, pays, deletes, invites or signs out. Where a login needs a person, such as an SMS code or a captcha, it stops and asks you.
Do I need to connect analytics?
No. The visual side works on its own. Connect GA4, PostHog, Mixpanel, Amplitude, Hotjar or Clarity, or add Corexi's snippet, and the behavioural side joins in. It is only scored when a source measured it.
How is this different from Hotjar or Lighthouse?
Hotjar records what people did, and someone has to watch. Lighthouse checks one public page's technical quality. Corexi signs in, walks every screen behind the login, reads it through five lenses, combines it with your analytics and hands you prioritised fixes, after every release.
Does it work on mobile apps?
Web apps are walked at the width each login is set to: desktop, or a phone for products people use on phones. A mobile app is added as its own product: built on a webview, it is walked on a phone; native, it is scored from the screens you upload, with the same findings and lenses.
What does it cost?
Plans depend on how many products you run it on and how often. The first 14 days are free, with no credit card, and we set it up with you on a 15-minute call, where we also go through pricing for your case.
From finding the fix to shipping it.
Today it shows what stands between people and your goal, and hands your team the fix. Next, it ships the fix itself, checks it with real users and keeps what works.
- v1.0Live now
Finds it, prices it, hands it over
- Signs in and walks every screen, at desktop or phone width
- Reads each screen through five lenses and weighs it against your analytics
- Prices every finding against the goal you set
- Hands the fix to Jira, GitHub, Slack and your editor over MCP
- v1.xNext
Goes further on its own
- Walks native iOS and Android apps, not only the screens you upload
- Opens the fix pull request itself, for your team to review
- Scores AI readiness: whether AI agents and assistants can use your product, from finding the button to finishing the checkout
- v2.0Where it is going
Self-driving UX
- Ships the fix behind a feature flag, to a slice of your users
- Measures it against your goal with real people
- Keeps it or rolls it back, and tells you which and why
LED-214 · Send invoiceflag · 10%Invoices sent, first weekWithout the fix61%With the fix68%Kept, and rolled out to everyone
v1.0 is what runs today. The later versions are our plan, shown without dates.
Then it goes round again.
Give it your product's URL. We set up the login with you once; from then on it goes round on its own, after every release.