AI QA experts · for engineering & QA leaders

AI writes more code.Testing has to keep up.

Our AI QA experts help you keep up now—and make your own testing AI-first for good. Start with a 30-minute call.

Keep up now

Add AI QA capacity

Our AI QA experts test your changes alongside your team.

  • Works in your tools
  • Expert-verified findings
  • On for a release, off after
Book a capacity call →No transformation project required.
Level up for good

Go AI-first

Our experts use the CARBON framework to transform your testing in 30 days.

  • Keep what works, convert the rest
  • AI agents take the repetition
  • Your team owns the result
🤖 Yes, even our AI transformation is powered by AI.
Book a transformation call →See the full 30-day process →
The 92% AI can do, and the 8% it missesVenn diagram. A large AI circle covers 92% of findings, including 13% shared with QA experts. A smaller QA-expert circle extends beyond it; the expert-only crescent is the 8% AI misses. 92% AI can do it 13% together 8% AI misses AI QA experts AI testing
Why AI QA experts

We know the 92% AI can do.
And the 8% it misses.

92%AI finds on its own or with expert steering. We run it at scale.
8%Only experts catch: security, look & feel, business rules, edge cases.
Our AI QA experts review all AI output. Our Human Review tool makes that fast: we sample by risk, with statistical confidence.

Working split from IcebergQA research (79% AI-only, 13% shared, 8% expert-only). Varies by project.

The people

AI-first testers.
Not testers learning AI.

The 8% needs people. Every IcebergQA engagement is run by senior QA experts who work with AI all day—and who helped build the CARBON tools they use.

Claude since day oneOur experts have used Claude and AI coding agents in their daily testing from the start.
Co-designed CARBONThey wrote the requirements for Eval², Software Eval, and Human Review—and test every release.
Review every AI resultNothing AI finds reaches you without an expert’s judgment.
Teach, not just doThey work beside your team and leave the skills behind.

Jason Arbon

Co-founder · AI testing
  • Co-author of How Google Tests Software
  • Tested at Google and Microsoft; founded test.ai, an early AI testing company
  • Builds CARBON at testers.ai; author of How AI Tests Software

Philip Lew

Co-founder · QA services & metrics
  • Has run a software testing services company since 2006
  • Builds quality metrics for clients and teaches metrics courses
  • Writes the IcebergQA blog on confidence and accountability
Read Phil’s latest →
Built by the people who use it

They don’t just use AI tools.
They built them.

Most QA vendors buy their AI. Our experts defined what CARBON had to do—from how Human Review samples results to what Software Eval measures—then used it on real client work and fed back what broke.

So when they transform your testing, they’re handing you tools they trust with their own names.

A day with an AI QA expert
  1. Agents runCARBON runs broad checks across your latest changes.
  2. TriageThe expert reviews every AI result in Human Review.
  3. The 8%Deep work AI can’t do: business rules, edge cases, look and feel.
  4. Release briefVerified findings, fix prompts, and a clear ship call.

Meet one of them.

On the first call you talk to an AI QA expert, matched to your stack—not a sales rep.

Book a call with an expert →
How it works

Start with capacity. Grow into AI-first.

Click a step to see what happens.

Add capacity: steps 1–3Go AI-first: adds step 4See the full process →
Our software

Experts with real tools.
Not just prompts.

We built the software our AI QA experts use every day—and leave it running for your team. Want a tour? We’ll show it live on the call.

CARBON Studio

Our quality workspace. Jay, our AI test manager, runs the checks while our experts steer—evidence, findings, and coverage by domain in one place.

  • AI agents run functional, UI, API, security, accessibility, and persona checks
  • Every finding links to captured evidence
  • Experts steer the agents toward your highest risks

Product screenshots shown with sample data.

From the IcebergQA blog

Thinking on AI and testing.

Book a call

Let’s talk for 30 minutes.

Tell us where you are. An AI QA expert will email you to pick a time.

  • 1
    30 minutes with an AI QA expert—not a sales script.
  • 2
    We look at your backlog, test suite, or AI features with you.
  • 3
    You leave with a recommended next step: capacity, transformation, or neither.
What do you need?

We’ll use your details only to respond. Don’t include passwords or customer data.

Prefer email? hello@icebergqa.com

Book a 30-min call →