Hey!

Welcome back to another week of musings. As you're reading this, we're observing Martin Luther King Jr. Day in the US, so I'm probably enjoying a mimosa while watching people go by.

This past week really felt like the year had fully started. So, a whirlwind of meetings, catching up, and setting goals.

Was this forwarded to you? You can subscribe here!

Things I enjoyed in the past week

• 10 Book Pairings to Expand the Way You Think. I recommend this Substack, but this post is interesting for finding some new books.
• I recently re-started The Staff Engineer's Path because I was wondering I misunderstood some pieces.

Lately, with more and more conversations about how to better leverage AI at work and within teams, ideas keep floating through my mind.

Curiously, we've been having these conversations since the rise in popularity of ChatGPT, then Claude, and now with GPT-5.2 and Opus-4.5; it feels like the originals were aeons ago. These are making the idea of not doing that much code by hand more and more real.

These new models, along with things like sub-agents, skills, commands, etc I think we're going to see a decline in "vibe coding", and the engineering rigor will be more visible.

Seeing things like Vercel's React Best Practices makes me think teams can now encapsulate many of their idiosyncrasies as shareable skills. You can be producing and reviewing code up to a certain standard, even from new hires.

I also came across this idea from The Pragmatic Engineer about what happens when AI writes most of the code, and it planted the idea of a declining value of certain aspects, as engineers can now jump not only to unknown code bases but also to unknown languages. I think having deep knowledge of a language or runtime will always be valuable at the edges, where the rubber meets the road, if you will.

What I'm noticing, and what is painful for me and the teams I work with, is that engineering cultures that document more, or have more established processes like RFCs, HLDs, or other types of documentation, will get more leverage out of them. If your tickets and test cases are well written, you could probably kick off a pull request, as they're accepted for work during the sprint.

If anything, I keep feeling I'm late to this thing. But realistically, I'm not. Anyways, that's all for today! short and sweet.

Your turn!

Have you considered leveraging AI more at work? Have you thought what it would look like if AI wrote all the code? Let me know your thoughts by replying to this email!

Happy coding!