If you could rebuild a significant part of your work that was created before AI, what would you do differently?

One thing I'd rebuild differently is a validation tool I wrote for a small internal data flow. At the time, I spent a lot of effort reading old examples, copying patterns, and manually writing very similar checks, so I got the happy path working but found edge cases late. If I rebuilt it now, I'd use AI early to generate likely invalid inputs, draft repetitive validation logic, and show me a few implementation options so I could compare them faster. I would still test the final behavior myself, because the mistakes I made were mostly around assumptions, not syntax. The biggest difference is that I'd spend less time producing the first version and more time making sure the error handling is actually useful. That would probably have reduced the rework I had during testing.

If I could rebuild the scheduling dashboard I worked on for our operations team, I would spend a lot more time on the user flow before writing the first line of code. I built it by adding features as people requested them, so it worked, but the layout got cluttered and a few buttons ended up in places that were confusing for new users. If I had AI back then, I would use it to quickly draft a few interface options, rewrite plain-language instructions, and summarize feedback from early notes so I could spot patterns faster. I’d still want to sit with the people who used it every day, because the main lesson for me was that the best answer was less about code and more about making the tool easy to trust. What I’d do differently is use AI to move faster on the first draft, then spend my time on clarity and simplicity instead of just adding more features.