Building AI products

The first thing I want from AI doing product marketing is the ability to say "I don’t know." If it can’t verify a competitor change, that uncertainty belongs in the answer. If a recommendation needs evidence it doesn’t have, the system should stop there. Friday at 10 AM PT I’m

The supermarket shelf used to hold maybe 1,000 SKUs. Now it can hold 10,000, because AI makes it just as easy to build the wrong thing as the right one. Brian (Aha!) argues this is exactly where product managers earn their keep: strategy and discovery decide what deserves a spot on the shelf…

“If we truly believe that AI is changing how we work, then ICs should get paid more than managers, because an IC can drive more impact than a manager could in the pre-AI system.” — @ElenaVerna

If you use AI for competitor research, steal this rule. Every claim gets a date. Last week Claude answered a simple question about Linear’s competitors using an old source as if it were recent. It also stated a guess as fact. The skill forced the answer to show the dates,

If a customer conversation has ever changed how you explain the product, you’ve already done product marketing. Friday at 10 AM PT I’m taking one B2B product through 16 jobs like that with AI. I’ll run the same work with and without the matching skill and see which version I’d

This is what separates teams that accelerate with AI from those that get stuck. You don't find perfect data lying around. You create it using the deep product intuition you already have. /end Check out the whole episode with @vlaurenlee here:

.@thsottiaux (Head of ChatGPT & Codex) on the skills trending up and down in the AI era. Trending up: great taste, passion to build something that matters, knowing what good looks like. Trending down: typing fast.

The software of the future promises more power and flexibility for more imagination. We started with use-case software, moving to platform software, moving to all-purpose coding agents. Interface design is shifting rightward. A few things to keep in mind for designers: 1. Make the interface…

Teresa Torres wants to believe every product team does careful synthesis after each customer interview. The reality? It's hard, so teams give up and grab the two or three things they remember most. "We're not competing with making good synthesis faster. We're competing with making shallow synthesis…

I'm seeing a huge difference in performance (in a bad direction) across my evals moving from Sonnet 4.6 to Sonnet 5. Previously, my service ran at temperature 0. I'm wondering what prompt strategies people are using to constrain Sonnet 5 now that temperature is no longer an option.

AI can tell you why you lost a deal even when it has no idea. I asked Claude why Linear is losing deals to Jira. All it had was public web research. There were no buyer interviews, call notes, or deal records behind the answer. Claude still gave me reasons. With the win/loss

They're taking everything they know about commerce, customer behavior, and product strategy, and using that knowledge to systematically generate the training data they need. It transforms the data bottleneck from a passive waiting game into an active engineering challenge. /3

Product Hunt's doing something new today. Years ago we talked about adding community-driven changelog. Now with agents, the user request/bug report is the start of the prompt.