Stand up the team
Design the small, AI-augmented team that builds and validates a product. Name an owner for quality and user contact, and work out how generative AI fits across research, design, build, and test without turning into a liability.
Volume I · The method, remade for 2026
Ship in a day. Earn trust release by release.
Continuous improvement, with quality at the core
Kaizen-UX is Peter W. Szabo's method for making software when AI writes most of it: a small team, generative AI across the whole loop, and a real user every week. Codified in User Experience Mapping (2017); rebuilt now that shipping is the easy part.
Volume II · Provenance & particulars
2008Practicing UX since
2017Codified in the book
2People the method needs
1Improvement every cycle
Research and counsel for
- Samsung
- Amazon
- Flutter Entertainment
- DEWA
- HSBC
Volume III · The Thesis
On building things worth trusting
Generative AI can produce a passable product in a day. The hard part now is judgment: knowing whether the thing is correct, usable, and worth a person's trust, then improving it in a tight loop with real users. Kaizen-UX keeps the old conviction that the user's needs come before the business's goals, and adds a second: verify what AI produces and validate it with real people before anything ships. Improve a little each day and the gains compound.
“Build is cheap; judgment is scarce.”
Volume IV · Advisory
Three ways Peter helps
Each rests on the same conviction: when anyone can build, the team that keeps improving against real users outlasts the rest.
Put quality at the core
Move quality assurance and real-user testing from the end of the pipeline into every cycle. Evals for the AI-powered features, usability and dogfooding with real people, and accessibility that holds up under a screen reader.
Mainstay
Install the loop
Set up the Kaizen-UX cadence of measure and learn: small batches, one improvement every cycle, a pace the team can hold. You keep improving the product long after Peter's part is done.
Volume V · The Method
How the work runs
A loop borrowed from the factory floor and retooled for a team working alongside AI: see the real user, let AI build, verify before you ship, learn from what changed, grow the team as you go.
See · Build · Verify · Learn · Grow
Go and see, for real
Start with a real user, not the roadmap, and talk to one every week. AI can recruit, transcribe, and draft the synthesis; a human checks every insight against the recording.
Let AI build; you decide what's good
Generative AI writes most of the code and drafts most of the options. Reviewing what comes back is the human's job now; the bottleneck moved from typing to judgment.
Verify before you ship
Quality comes first: evals for anything AI-powered with no single right answer, tests and security scans by default, and real people trying it before it goes out, assistive-technology users included.
Measure, learn, compound
Ship small, measure what changed, and commit to one improvement each cycle. Small, compounding gains beat the big rewrite.
Grow the people
Use AI to widen what each person can do, and guard a sustainable pace; burnout is the tiny team's top failure mode. Continuous improvement is for the people as much as the product.
Volume VI · The Manifesto
The Kaizen-UX Manifesto
Five articles carry the spirit of the Agile Manifesto into the craft of user experience. Five more come from building with AI. We value the items on the left without dismissing those on the right.
Maps & communication over comprehensive documentation
People & interactions over processes & tools
Customer collaboration over contract negotiation
Responding to change over following a fixed plan
The needs of the user over the goals of the business
Verified evidence over confident assumptions
Real users over plausible output
Reviewing the work over generating more
The smallest thing that works over the speculative build
Growing people over faster output
The first five appeared in User Experience Mapping (Packt, 2017).
Volume VII · Questions
Questions, answered plainly
What is Kaizen-UX?
Peter W. Szabo's method for building software when AI does most of the making. It stands on three pillars: Kaizen (continuous improvement), user experience (real users first), and AI-amplified practice (AI across research, design, build, and test). The name carries the first two; quality assurance and real-user testing sit at its core.
Who is it for?
Founders and small teams building products with AI, and the product, design, and engineering leaders who want quality and real-user evidence woven into the work instead of bolted on at the end.
How is this different from “agile UX” or “vibe coding”?
It keeps agile's spirit and adds two rules it will not bend on: the user's needs come before business goals, and evidence comes before vibes. You verify what AI builds and validate it with real users before it ships. Vibe-code to explore, then apply engineering discipline before you deliver.
How do engagements work?
Most begin with a working session on the problem in front of you, or a short audit of how you build, verify, and learn. From there, teams keep Peter on the measure-and-learn cadence, embedded or advisory, remote or on-site. The method holds whether you are a two-person founding team or a product org.
Who is Peter W. Szabo?
A UX practitioner since 2008 and author of User Experience Mapping (Packt, 2017), the book where Kaizen-UX began. He founded one of Romania's first UX agencies and has led UX teams as a director; his research and advisory work has served Samsung, Amazon, Flutter Entertainment, DEWA, and HSBC.
Volume VIII · An Invitation
Start the work
Standing up a tiny team, making quality the default, or building the loop from first principles: the first improvement is a conversation.