Principles to not be an AI cowboy
The key principles from Agile teams that AI accelerates but doesn't replace. Focus on incremental development, rigorous testing, and solid standards for architecture, security, and quality. Embrace the "boring" stuff - it's what prevents disaster and, with AI, can actually be a lot of fun!
Some lessons learned from my cowboy crusade in the vibe coding seat were the same things known by Agile product and dev teams. AI doesn’t eliminate these principles; it just makes them much faster to implement.
(For non-tech folks, don’t freak out. AI can help with all of these, too.)
- Be incremental in product development.
- Get an MVP with a core set of features
- Beat on it, refine it, tweak it
- Test, test, test
- Get that small set of features hardened
- Test some more
- Release it, get real-world feedback, and determine your roadmap from there. This separates an actual product from just having a hobby.
- Define and enforce product development standards.
- Architecture, security, UX, QA, Data, … - define these and let AI help you
- Be rigorous in your technical development
- Conduct code reviews, refactor, refine, and polish
- Implement peer reviews
- Build like a cool Fonzy and not like a Tazmania devil
“Boring!” you’re thinking.
I get it. As children, we all wanted ice cream instead of vegetables. As adults, though, we mature to actually ENJOY healthier foods that avoid database corruption, API errors, cloud budget explosions, and SQL injections.
I’m mixing metaphors. The punchline is that all of this “boring” stuff is important and can actually be a lot of fun with AI. The next article will get your hands back on the keyboard with some fun, healthy activities sure to entertain.
