Now in public beta — Powered by Claude Vision AI

Ship pixel-perfect UI.
Every single time.

AI vision catches visual bugs, layout regressions, and accessibility issues before your users do. Drop into any CI/CD pipeline in 5 minutes.

Start Free → See How It Works

No credit card required · 15 free analyses · 7-day trial

Why teams choose iVisionQA

AI that sees your UI
like a real user does.

Not pixel-diffing. Actual visual intelligence that understands layout, design systems, and user experience.

AI Visual Review

Claude Vision AI analyzes every page for layout issues, typography problems, spacing inconsistencies, and color contrast — like a senior designer reviewing every deploy.

Smart Regression Detection

Goes beyond pixel-diffing. AI understands intentional changes vs. bugs. Dynamic data, timestamps, and avatars won't trigger false positives.

Accessibility Auditing

Automated WCAG 2.1 AA scanning with axe-core, enhanced by AI analysis of visual accessibility issues automated tools miss.

Cross-Device Testing

Test across Chrome, Firefox, Safari, and mobile viewports. AI verifies your responsive design actually works on every screen size.

CI/CD Native

Drop-in configs for GitLab CI, GitHub Actions, Azure DevOps, Jenkins, and CircleCI. Tests run on every push and deploy.

Quality Scoring

Every page gets a 0–100 quality score with categorized issues by severity. Track quality over time with trend dashboards.

See what pixel-diffing
can't see.

Traditional tools flag every pixel change. iVisionQA understands your UI and tells you exactly what broke and why.

BASELINEv2.4.1
CURRENTv2.5.0 — 3 issues
AI ANALYSIS
CRITICALPrimary CTA button width increased 75% — now overflows container on mobile viewports. Likely caused by unbounded text content.
WARNINGSecond card border changed from default to red. If intentional (error state), confirm contrast meets WCAG AA (currently 3.2:1, needs 4.5:1).
INFOCard content text updated — appears to be a data change, not a layout regression. Marked as allowed dynamic content.

Testing that works
while you sleep.

iVisionQA agents run autonomously — monitoring your production UI and fixing issues before you even see them.

Continuous Monitoring
Scout Agent

Runs on a schedule — hourly, daily, or custom. Crawls your production site, takes screenshots, runs AI analysis, and compares against baselines. Catches regressions between deploys that CI/CD misses.

  • Scheduled crawls — hourly, daily, or cron
  • Catches CDN issues, third-party script breakage, CMS changes
  • Instant Slack / Teams / PagerDuty alerts
  • Auto-triggers Fix Agent when issues are found
  • Quality trend tracking over time
Auto-Remediation
Fix Agent

Doesn't just report issues — opens a pull request with the fix. AI analyzes the visual bug, generates a CSS/HTML patch, creates a branch, and opens the PR. You review and merge.

  • Auto-generates fix PRs from detected issues
  • AI-authored CSS, HTML, and layout patches
  • Full diff with before/after screenshots
  • Works with GitHub and GitLab
  • Human-in-the-loop — you always review before merge

Three commands.
Infinite quality.

AI-powered UI testing in under 5 minutes. No complex setup.

In your CI pipeline

Install

One package. Zero config files to manually create.

npm install ivisionqa

Initialize

Auto-discovers pages, generates test configs. Point it at your app.

npx ivisionqa init --scan https://yourapp.com

Ship

Runs in CI/CD. Every push gets AI-reviewed. Regressions caught.

npx ivisionqa run --ai --notify slack
In your AI agent (MCP)

Add to your MCP config

Works with Claude Code, Claude Desktop, Cursor, or any MCP-capable agent. 

{
  "mcpServers": {
    "ivisionqa": {
      "command": "npx",
      "args": ["-y", "@haivio/ivisionqa-mcp"],
      "env": {
        "IVISIONQA_API_KEY": "ivq_your_key"
      }
    }
  }
}

Your agent gains four tools

analyze_screenshot scores any screenshot for layout, typography, spacing, and accessibility issues. compare_screenshots catches visual regressions against a baseline. get_usage checks your quota. ask_support answers questions in-session.

Close the loop

Your agent builds a UI, screenshots it, calls analyze_screenshot, reads the issue list, fixes, and re-checks.

Use it your way

MCP server npx -y @haivio/ivisionqa-mcp
CLI npx ivisionqa scan http://localhost:3000
Playwright fixtures Drop-in test helpers
CI/CD configs GitLab, GitHub Actions, Azure DevOps
REST API POST /v1/analyze — docs

Try free. Scale as you grow.

Every plan includes AI-powered analysis. No per-seat pricing.

Trial
$0
15 analyses, 7 days, no credit card.
  • 15 AI analyses
  • 1 project
  • Chrome testing
  • 7-day trial period
  • Community support
Start Trial
Starter
$50/mo
or $500/yr — 2 months free
For growing teams.
  • 300 AI analyses / month
  • 3 projects
  • Chrome + Firefox
  • 30-day report retention
  • GitLab & GitHub CI configs
  • Email support
Get Starter
Pro
$100/mo
or $1,000/yr — 2 months free
For teams shipping production apps.
  • 2,000 AI analyses / month
  • Unlimited projects
  • All browsers + mobile
  • 90-day report retention
  • Scan-run support
  • Slack & Teams notifications
  • Visual regression baselines
  • Quality trend dashboard
  • Priority email support
Start Pro
BYOK
$200/mo
or $2,000/yr — 2 months free
Platform fee. Bring your own Anthropic key.
  • Unlimited AI analyses
  • Your Anthropic API key
  • Per-request model choice
  • Everything in Pro
  • Scout + Fix agents — Coming Soon v2.1
  • No platform rate limits
  • Full API access
  • Priority support
Get BYOK
Enterprise
Custom
For orgs with compliance needs.
  • Unlimited AI analyses
  • Unlimited projects
  • Scout Agent — custom schedules
  • Fix Agent — auto-generated PRs
  • SSO / SAML authentication
  • Self-hosted runner option
  • Custom AI strictness policies
  • SOC 2 compliance reports
  • Dedicated support engineer
  • SLA guarantees
Contact Sales

Fits your stack.

Works with the tools you already use. No vendor lock-in.

GitLab CI/CD
GitHub Actions
Azure DevOps
Jenkins
Slack
Microsoft Teams
Playwright
Selenium
Cypress
Kubernetes
Docker

Questions?

Can't find your answer? Email info@ivisionqa.dev.

Those tools do pixel-level comparison. iVisionQA uses Claude Vision AI, which understands your UI semantically. It knows the difference between a bug and dynamic content — zero false positives from timestamps, user data, or animations. And we're 10x cheaper.
Bring Your Own Key. BYOK is its own premium tier at $200/mo (or $2,000/yr) platform fee. You use your own Anthropic API key for AI compute; we charge for the platform (gateway, routing, quota enforcement, logging, dashboard, and support). BYOK is above Pro, not a discount.
Under 5 minutes. Install the npm package, run the init command pointing at your app URL, and run your first test. Init auto-discovers your pages and generates configs.
Yes. Drop-in configs for GitLab CI, GitHub Actions, Azure DevOps, Jenkins, and CircleCI. iVisionQA is an npm package — runs anywhere Node.js runs.
Screenshots are analyzed by Claude Vision AI. Your data is never used for training. On Enterprise plans, self-hosted runners keep screenshots in your network.
Yes. iVisionQA Enterprise supports self-hosted deployment and air-gapped environments. Self-hosted runners keep all screenshots and data within your own network.
Yes. Add the iVisionQA MCP server to Claude Code, Claude Desktop, Cursor, or any MCP-capable agent. Your agent gets four tools: analyze_screenshot, compare_screenshots, get_usage, and ask_support. It can build UI, screenshot it, score it, fix issues, and re-check — fully autonomous visual QA.

Stop shipping
broken UIs.

Start with a free 7-day trial — 15 analyses, no credit card. See your first AI quality score in under 5 minutes.

Start Free → Talk to Sales