How AI is changing product management (and what to ask your team about it)
Hey friends,
Welcome to another belated edition of what's new on the blog. It's been a bit of a slow month since I've been traveling quite a bit, but I published a new post this morning so I thought now would be a good time to send you your beloved link list as well. Enjoy, and have a good weekend!
Rian
Full Posts
How AI is changing product management (and what to ask your team about it)
Building got cheap, and deciding what to build got expensive. That's one of the few thing everyone agrees on. I read the people who shaped how PMs think about the job (Marty Cagan, Melissa Perri, Teresa Torres, Rich Mironov, John Cutler, the Reforge team) alongside product leaders doing it right now at Anthropic, OpenAI, Netflix, Instagram, and Whatnot, then checked all of it against the 2025-2026 research from DORA, GitClear, and Perri's State of AI in Product 2026.
They also agree that teams are getting smaller and more senior, and that speed alone isn't converting to outcomes. The real disagreement is validation. Once you can ship in a day, do you still need to find out first whether people want the thing? Cat Wu at Anthropic says shipping is the test. Leah Tharin, who writes about product-led growth, says finding out is still the slow part, no matter how fast the code arrives.
I close with three questions for your own team: what breaks first as engineering gets faster, what a PRD (product requirements document) is still doing for you once the prototype is the spec, and who owns the evals, the repeatable tests that grade whether an AI feature's output is any good.
Link Posts
- Should You Use AI for a Task? Here's a Simple Way to Decide — Bruce Schneier's split between "work" tasks, where the job just has to get done and any assistance is fair game, and "gym" tasks, where the whole point is that you did the lifting yourself.
- In the Age of AI, Esther Perel's Relationship Counseling Is More Necessary Than Ever — Perel on why AI companions are so appealing (sycophantic, undemanding, frictionless), and why the people who turn to them are usually already isolated by the time they get there.
- Instead of Taking Your Job, A.I. Might Transform It — Early PCs shipped with BASIC so ordinary people could write their own software, and then VisiCalc, the first spreadsheet, buried that idea for decades. Cal Newport on the history, and what vibe coding brings back.
- AI enthusiasts are in a race against time, AI skeptics are in a race against entropy — Charity Majors on why the enthusiasts are right that the wins are real and the skeptics are right that the costs compound, and what it would take for the two camps to work the problem together instead of past each other.
- Social Media Is Now Parasocial Media — Platforms that used to be for talking with friends are now broadcast media where you have to work the algorithm just to be seen. danah boyd on how that happened, plus a plea to bring back Path, which solved this a decade ago.
Side project releases
- TL;DL v2.5.0 — Episode summaries now include their editorial deck and pull quote in the email newsletter and RSS feed, and transcription moved to OpenAI's gpt-transcribe model, which stops cutting off before the end of long episodes.
- discogs-mcp v3.4.0 — Your Discogs wantlist now works end to end. Add a release the moment you spot it, and clear it off once you finally track it down, without leaving your LLM client.