I Spent 300M Tokens So You Don't Have To
A test engineering manager's journey from skeptic to advocate, revealing the real value of AI in quality engineering through 300 million tokens of hands-on experience.
Most people dabble with AI. They copy a prompt from Twitter, paste it into a tool, skim the output, and either walk away saying "that's rubbish" or "wow, scary fast."
I didn't dabble. I went all in.
Over the past year, I spent more than 300 million tokens working inside Bolt's AI platform. Not to chase gimmicks. Not to "automate everything." I used it to test, break, rebuild, and really see whether this stuff can help engineers and testers in the real world.
The Context
I came at it from a specific angle: as a test engineering manager at Booking.com. My Manchester team is steeped in XP practices like TDD, mobbing, and pair programming. My role has always been about raising quality while cutting through noise.
So the question was not whether AI can code. We already know it can. The question was whether it makes our teams stronger.
The answer is yes. Just not for the reasons most people think.
The Do's and Don'ts
After 300 million tokens, here's what I would pass on to anyone starting out:
Do
- Use it for scaffolding, boilerplate, and prototypes.
- Plug it into your existing flows like TDD, CI/CD, and exploratory sessions.
- Let testers lead the AI conversation; their instincts keep everything honest.
Don't
- Skip reviews or compliance checks.
- Assume outputs are production-ready.
- Fall into hype or cynicism. Both waste time.
Why This Matters
"Testing has never been about catching bugs. That is surface-level thinking. The real purpose of testing is to accelerate learning."
Learning about systems. Learning about risks. Learning about how real people will collide with the product.
AI does not change that. What it changes is the starting point. Instead of losing hours untangling weak scaffolding, we can step straight into the meaningful work.
From Experiment to Framework
I have taken those lessons and turned them into something bigger: Viyb Testing.
Viyb Testing is not a buzzword or a shiny tool. It's a framework that combines what I learned from burning through 300 million tokens with proven agile practices.
Viyb Testing Framework
Disciplined AI for Confident Quality
Measure Twice, Cut Once: Accelerate Learning, Not Just Speed.
The Viyb Testing Mindset
AI is a "Super-Charged Junior Dev"
Fast, tireless, but needs guidance. Don't worship, don't dismiss.
Shift from Firefighting to Discovery
Move beyond patching bugs.
Focus on Confidence, Not Just Speed
Quality scaffolding empowers deeper testing.
Ready to transform your testing with disciplined AI?
Whether you're looking for a speaker, a consultant, or a new team to tackle your biggest quality challenges, reach out if you'd like to learn more.