Scott Belsky on product sense, AI, and building through the messy middle

Executive overview

Most teams obsess over features while neglecting the first 30 seconds of their product experience — the moment when users are lazy, vain, and selfish. Get them through that window and you earn the right to a lasting relationship. Strip features ruthlessly: Behance grew faster every time it killed something.

AI won't replace product roles — it will expand the surface area of possibilities each person can explore. The PM, designer, and analyst each become more cross-functional, not redundant.

The founders who survive the messy middle are those who have more conviction now than when they started — not less.

Building product sense

• Empathy for the customer's problem beats passion for your proposed solution — passion can be 30 degrees off.
• Develop empathy shoulder-to-shoulder with customers in their full daily context, not just while using your product.
• Users in the first 30 seconds are lazy, vain, and selfish — they want speed, to look good, and to feel successful immediately.
• Most teams spend the final mile of build time thinking about the first mile of customer experience; invert this.
• First-mile covers onboarding, orientation, and defaults — users need to always know how they got somewhere and how to get home.
• Customer cohorts change over time; early adopters forgive friction that pragmatist-stage customers won't — rethink onboarding repeatedly.
• Surprise and delight drives word of mouth: people talk about what a product does that they didn't expect, not what it does as expected.

The half rule: ruthless reduction

• At Behance's launch, more features meant more complexity — killing the tip exchange sent project publishing up; killing groups sent it up again.
• When a product focuses on one core action, that action runs at 10x velocity.
• Optimize for the problems you want to have — don't build features customers haven't yet cared enough to ask for.
• Every time you add something, consider what you can replace or remove.
• Removing Behance's portfolio color controls for 24 hours drew complaints; after 24 hours, silence — portfolios looked cleaner and core metrics improved.
• The best product leaders have a reductionist default: they anchor on the one thing they want users to do and treat everything else as a distraction.

What makes durable consumer products

• Lasting consumer products create new network effects or unlock latent capacity (e.g., Uber, Pinterest).
• Pinterest worked because it addressed consumer psychology differently — representing interests rather than inducing social anxiety.
• Pinterest also had an underlying business flywheel: driving traffic to pinned sources got those sites to add pin buttons, expanding distribution.
• Many recent consumer apps are just clever momentary interfaces — effective R&D for the platforms that already own distribution, who then copy the feature.
• As users become more technologically literate, they tolerate more cognitive load, making it easier for established platforms to absorb novel features.

AI and the future of product work

• AI moves teams from workflow to flow — removing the friction between an idea and its development.
• Like engineers who became more productive but more in-demand, higher human ingenuity per person may mean companies want more people, not fewer.
• ChatGPT mines the center of what already exists; it won't find the edges that become the center — human ingenuity still does that.
• AI's primary product impact: expanding the surface area of possibilities each person can explore in less time.
• A Hollywood director uses ChatGPT not to write scripts but to generate five scenarios he doesn't want — useful negative data.
• AI will collapse functional silos: PMs will do more design, data, and engineering without needing specialists for every query.
• Personalization is the third wave — apps will increasingly meet each user where they are, replacing generic onboarding with tailored experiences.
• Advice for PMs: play. Use every tool available. Risk of experience is getting stuck in your ways.

What Scott looks for in founders and products

• Values founders who listen, want to shake things up, and are mission-motivated over money-motivated.
• Red flag: promoters who sugarcoat — great founders are optimistic about the future but pragmatic and somewhat pessimistic about the present.
• Wants founders who can name exactly what's not working, what keeps them up at night — those are real people who can partner through the messy middle.
• On product: looks for a clear object model — every screen should answer: how did I get here, what do I do now, what do I do next?
• A clean object model signals the team values navigability as a core principle; its absence is a red flag.

Surviving the messy middle

• The middle is messy because it is full of lows — you must endure anonymity, uncertainty, and anxiety.
• Human biology demands constant reward signals; building something takes years, which works against our instincts.
• Short-circuit the reward system with mutually agreed micro-milestones celebrated even when they don't matter in the grand scheme.
• As a product leader, merchandise progress — teams in the back seat with blacked-out windows will go stir crazy without a narrative.
• The quit test: ask whether you have more or less conviction now than when you started. More conviction despite failures = messy middle, keep going. Less conviction = quit and pivot.
• Don't make a bold decision on a bad day, but if conviction generally dissipates, be open to other options.

Resourcefulness over resources

• The last decade's default was throwing resources at every problem; the current constraint forces a different muscle.
• Resourcefulness stays with you like muscle; resources are like carbs — consumed and gone.
• The best teams in constrained environments build a refactoring habit that compounds over time.
• Exceptions are the rule in truly transformative work — nothing extraordinary is achieved through ordinary means.