Samplers fundamentally changed how hip hop artists make music in the 80s. Artists no longer needed formal training to create something great. They could sample existing sounds, rearrange them, and build something entirely new.
This story stuck with me when I first came across this podcast. It got me thinking: what would it take for a product manager to actually build software in the age of AI-assisted tools?
The challenge
After rapidly building a few side projects with AI coding assistants, I felt empowered to make a small software change at work.
A customer had shared feedback that not having units listed in the drug file was confusing. For example, does a total quantity of “5” for Zarxio mean 5 mL or 5 syringes? To clarify this field, I needed to add a unit after the total quantity.
Sounds simple enough, right?
What actually happened
My change went live in production today. The actual code change took 5 minutes. But everything else, including the setup, testing, seeding data, linting, and increasing code coverage took days.
What I learned
1. Small customer delighters matter
Features like this often fall between priorities. They’re not critical enough to make the roadmap, but they’re exactly what makes software feel thoughtful. Shipping them matters to our customers.
2. Building in corporate production is way harder than side projects
My engineering colleague patiently guided me through the dev setup. Then came with many of the hurdles: confirming PR titles to standards, fixing linter errors in the existing code, and satisfying the coverage check that demanded tests for every change, even superficial inting fixes. I ended up writing tests for the existing code just to get the PR through.
Testing took another big chunk of time. I had to seed sensible data to verify the change actually worked.
3. We’re all product builders
Whether the role is product manager, engineer, or designer, we’re all building the product together. In the age of AI-assisted tool, the barrier has never been lower.
Samplers didn’t replace musicians. Rather, they expanded who could participate in music creation. AI coding tools might be doing the same for software. It’s time for PMs to be curious to try.