đ§ The Thinking Trap / PM Snacks 93
Plan vs Strategy ⢠The Modern AI PM ⢠Talk to Fewer Customers ⢠Being Right Is a Trap ⢠Think First, AI Second
Hey everyone,
I noticed something fascinating this month: everyoneâs talking about speed â ship faster, iterate faster, prompt faster. But the sharpest thinkers I follow are all pointing at the same blind spot: weâre optimizing for velocity while neglecting the quality of our judgment.
In an age where AI can generate anything in seconds, the bottleneck has shifted from execution to thinking. Double down on judgment, or risk becoming noise.
Please support this newsletter by sharing it with your colleagues or liking it (tap on the đ).
Enjoy!
PS: sorry for being silent last dec, but after +10y without coding, I shipped my first solo app, entirely coded with codex and claude code in a month. Weâre entering a new world!
đ Snacks
#1
A Plan Is Not a Strategy ⢠June 2022 ⢠9 min watch ⢠#strategy
Roger Martin's litmus test: if your "strategy" doesn't make you nervous, it's probably just a plan. Real strategy means betting on an outcome you can't prove in advance, ideally capturing a theory, and differentiating from competitors.
Plans are lists of activities with no internal coherence. âImprove customer experience, open a new plant, launch a talent programâ, they all sound good, but they donât add up to winning.
Strategy requires a testable theory. Lay out what must be true about your market, your customers, your competition. Tweak when your assumptions don't hold.
Escaping the trap requires confronting uncertainty and making explicit choices about what you will and wonât do, and being willing to fail.
Next time you review your roadmap, ask: does this guide us through unknowns, or just make us feel busy?
#2
The Modern AI PM in the Age of Agents ⢠Jan 2026 ⢠7 min read ⢠#AI #execution
The PMâs role is shifting from translation to intent formulation. When AI agents can take a well-formed problem and produce working code, the spec is becoming the product itself.
The bottleneck isnât implementation anymore. Itâs knowing what to build. Problem shaping, taking an ambiguous pain point and making it clear enough for agents to act, is now THE skill.
Context curation is the new craft: maintain context docs that feed agents before starting any project. Include real user quotes, constraints, whatâs already failed, and how youâll measure success.
Taste and evaluation matter more than ever. Agents will confidently produce things that look correct but miss the point. You need to develop judgment.
Whatâs interesting to me is that this echoes what we discussed in PM Snacks #92 about the Super IC, but goes deeper into the mechanics.
#3
Sometimes, You Should Talk to Fewer Customers ⢠Jan 2026 ⢠3 min read ⢠#mindset
You need to understand that every activity that a big company & its CEO encourages â answer support tickets, talk to customers, respond immediately to email/Slack, etc. â is intended to push this now-large companyâs average & below-average employees to perform better, to get them to meet the now-lower bar of what the company expects from its median employee.
Shreyas Doshi delivers a contrarian take: relentless customer contact is a corporate best practice designed to raise the floor, not the ceiling.
If youâre already in the top 10%, you need to break the rules that were designed for the other 90%. The real work is quality of thinking, not volume of activity.
High-impact product work is about unique insights, bold bets, and protecting yourself from noise, not inbox zero or N customer calls per week. Ask: is this the best use of my time?
#4
Being Right Is a Trap ⢠Dec 2025 ⢠6 min read ⢠#AI #mindset
Hilary Gridley exposes a new paradox: sharp judgment in âAI worldâ can actually slow you down. What youâre right about today will probably be obsolete tomorrow. So make sure that you discard a tool or capability, you make sure to revisit it a few months later. For example AI-generated slides are bad quality (not true anymore with NotebookLM).
Said another way, this is one more reason why the best product builders should hold their convictions lightly and systematize humility.
#5
Most clicked link 1y ago.
The Lindy Standard of Performance ⢠Jan 2024 ⢠12 min read ⢠#execution
Hereâs the clearest articulation Iâve read on the mindsets, and tactics that drive outstanding execution. Starting with 2 questions to be sure that youâre working on the right thing:
1/ What must urgently be fixed in order to bridge our âperformance gap?â
2/ What must importantly be built in order to bridge our âopportunity gap?â
These two terms were coined by HBS professor Linda Hill. In a nutshell, bridging your performance gap means taking care of your existing business; if youâre Apple in 2000, your performance gap is the Mac. How do you keep taking market share in the PC industry?Bridging your opportunity gap means pursuing the things your business could become. For Apple in 2000, this was the iPod. You never run out of performance gap to bridge, but you must still find a way to bridge the opportunity gap â because if you donât do it, someone else will, and it will become your performance gap.
đď¸ Recently saved
Links worth reading that I saved, but did not highlight:
- Karpathy on X: LLMs arenât people, theyâre simulators. Stop asking âwhat do you think?â and start asking âwhat would experts say?â
- Ramp ships a major feature every day, hereâs their 7-point checklist for balancing speed with quality
- Still embracing speed, Sunil Pai's manifesto (cloudfare): "Ship in 6 weeks or die trying"
- Lenny & Karri Saarinen on the disappearing middle of software work: directing agent work is the new craft. Itâs exactly how I felt developing my mobile app.
â
Thank you for reading this far.
Until next time!
Olivier
Share â Please help me promote this newsletter by sharing it with colleagues (tap the refer button) or consider liking it (tap on the đ, it helps me a lot)!
About â Productverse is curated by Olivier Courtois (15y+ in product, Fractional CPO, coach & advisor). Each issue features handpicked links to help you become a better product maker.