0:00
/
0:00

Self-documenting software

Why AI levels the playing field for product teams and documentation

A meme in the startup world is how good Stripe’s documentation is.

Frankly, it is. It’s that good.

For the rest of us mere mortals, we can only aspire to have documentation that good.

Half as good, and we’d be happy.

Why is documentation such a pain point for product teams?

  1. Documentation is really hard. Writing good documentation is a skill, and the time and effort required to produce good docs is very high.

  2. Most product teams move faster than their documentation can keep up. Outdated documentation is frankly useless documentation.

I think AI levels the playing field somewhat however.

I’ve been playing around with this idea of “Self-Documenting Software” in my head for a while, and I think it’s kind of a big deal.

The idea that product and engineering teams can be consistently and constantly upgrading and updating documentation, in real time, as the product evolves and grows.

I’ve been using a combination of Claude Code and the Playwright MCP to experiment with just that.

Side note: I wrote about this combo the other day in the context of QA and testing, I really think it’s a killer duo that opens up a BUNCH of interesting use cases for product teams…

Anyways, the flow is something like: Read the docs, use the product, then update the docs.

In early testing, I’ve found it to be remarkably useful.

Thanks for reading! Subscribe for free to receive new posts and support my work.

Discussion about this video

User's avatar

Ready for more?