AI-Assisted Development Means Test-First Development #
Published on Aug 20, 2026 (updated Aug 22, 2026), filed under development, ai. (Share this post, e.g., on Mastodon.)
If you’re developing using AI, you need to start from tests.
If there’s one lesson I learned from developing, maintaining, and optimizing using all sorts of models—something I missed calling out in previous tips!—, then it’s this. Start from tests.
The way you do this—and you may need to be explicit as I’ve seen many models slap on tests after the fact—is by asking to first write tests for what you create or fix—and then develop the feature or implement the fix. Rinse and repeat.
Starting with tests protects your project:
- You confirm that the respective feature is missing or broken (and it’s recommended that you review and run the testsor, at least, take samples). - You put a guardrail in place so that any AI instance doesn’t veer off—any work it does should lead to developing or fixing the feature.
- You confirm that the feature was indeed built or fixed.
- It keeps yourself centered on the problem.
Now, I’m frontend developer first, web developer second, programmer last, and therefore look at some processes from an off-position. Still, working with AI has proven to me the value of test-first development. If you’re working with AI, and aren’t a strong programmer, try starting with tests, too. While you will need to get used to testing, this is easier and safer than “just shipping.”
Update (August 22, 2026)
The first version of this post linked the approach to test-driven development. Readers on Mastodon—thanks Ted M. Young and Benjamin Eskola—pointed out that this would not be accurate, that “test first” would be more appropriate. I do understand there to be a connection between test-first and TDD, but with my background as mentioned, I appreciate the feedback—and hence updated the post. “Test first” does well match the spirit of what I learned to be effective.
About Me #
I’m Jens (long: Jens Oliver Meiert), and I’m an engineering lead, guerrilla philosopher, and indie publisher. I’ve worked as a technical lead and engineering manager at various companies (e.g., Google); I’m an active web and tool developer (code optimization, digital defense), a contributor to web standards (like HTML, CSS, WCAG), and a book author (O’Reilly, Frontend Dogma).
I love trying things—in web development and engineering management, but also in politics and philosophy, where I hold to one idea in particular: that we can only be well if we take good care of everyone. Here on meiert.com I talk about some of my perspectives and experiences. (Interpret charitably but be critical—and share feedback and advice that makes my work better.)