From Tester to AI Operator: 20 Months of Shipping with AI Agents
An experience talk about shipping an iOS app in 20 months by directing AI agents, and why the skills that mattered were tester skills.
By Michał (Mike) Buczko, quality and delivery leader at Carrier/Viessmann and founder of BowSmith.
I started with zero iOS knowledge and one rule: I direct, AI agents build. Twenty months later the app, BowSmith, is in the App Store. This talk argues that the skills which made that possible were tester skills, and that a tester is closer to being a good AI operator than most people assume.
What the talk covers
- The starting point and the rule: no Swift, no iOS background, and no writing the code myself.
- The first setup: 32 role-based agents, modelled on the team I would have hired. Only 17 survived, and today six to eight remain, built around skills instead of roles.
- How it runs today: I talk to one AI coordinator, which does the work itself or through subagents, with a review at each handover.
- Failure patterns I ran into: tests that matched the code and not the requirement, a model approving its own work, and agents acting outside their brief.
- How I caught those failures, and what each one changed in the setup.
- Why the scarce skill was structuring work so the result can be trusted, and not writing code.
Who it is for
All levels. Testers and quality engineers who want to know where their skills fit when AI writes the code, and developers or managers who are setting up their first agent workflow. You do not need iOS knowledge.
What you take away
- A working setup you can adapt: one coordinator and a few skill-based subagents, with review at each handover.
- A list of failure patterns and how to catch each one.
- An argument for why testing skills are the core skills of an AI operator.
Where I have given it
New for 2027 — available for conferences and company events.
The talk draws on the same build as my 90-minute session at CAST 2026 (recap). Background: from 17 role-based agents to task-based Agent Teams and what a room of testers did to the thesis.
Slides and recordings
This talk is new, so there is no recording of it yet. Slides are available to organisers on request. Recordings of my earlier talks are on the Speaking page.
Format and practical details
- Length: 30–45 minutes plus Q&A, adjustable to the slot.
- Language: English or Polish. Polish title: "Od testera do operatora AI: 20 miesięcy dostarczania z agentami AI"
- Level: all levels.
- Room and tech: a projector or screen to connect to.
- Formats: conference talk or company event.
Related talks
- From Tester to AI Operator: Building the Career Path Before You Need It
- From Idea to Working Product: How AI-DLC Unlocks Human Creativity (CAST 2026)
- Is AI the future of Software Delivery: Can it replace a Feature team?
Book this talk
To book this talk, start at the Speaking page, book a 30-minute call or message me on LinkedIn.
