What four guests on Lenny's Podcast — spanning engineering, leadership, design, and growth — revealed about how products get built now.
Four conversations. Nearly six hours. One signal: the rules for building products just got rewritten. An engineer who hasn't touched his own code in months. A CEO coach who says nobody's coming to save you. A design leader declaring the entire design process dead. A founder who turned "the worst idea ever heard" into a $2B company.
Here's the best of what they said — and why it matters now.
100% of Boris Cherny's code is written by AI. Not a talking point — his actual daily workflow. Five agents running in parallel, shipping 10–20 pull requests a day, no manual edits since November. Productivity per engineer at Anthropic is up 200%.
Claude Code isn't autocomplete. It's evolving into a genuine co-worker — one that reads bug reports, analyzes telemetry, and proposes its own fixes. Boris compares the shift to the printing press: what was once elite craft is being democratized. Anyone can build software now.
The prediction that stopped the conversation: the title "software engineer" is going away. Replaced by "builder." By year's end, everyone codes, everyone's a PM. The old role boundaries collapse. Boris briefly left Anthropic for Cursor, came back in two weeks — not for the tech, but for the mission.
"I have never enjoyed coding as much as I do today because I don't have to deal with all the minutiae."
Boris ChernyBrian Halligan built HubSpot over 20 years. Now he coaches Sequoia's portfolio CEOs. His observation: at the "adults table" — companies past 100 employees — half your time should be recruiting. Everything else is downstream of the executive team you assemble.
And most of those hires won't work out. 50% of C-level executives churn within 18 months. His fix? Hire slow, fire fast. Do blind references, not gut checks. Send candidates the actual board deck — the ones who challenge it are keepers. The flatterers? Pass. Hire "spiky" people with exceptional strengths, not safe 3-out-of-4 generalists.
His framework for great CEOs: LOCKS — Lovable, Obsessed, Chip on shoulder, Knowledgeable, Student. The thread connecting all of them is what Brian calls "constructive dissatisfaction" — always improving, never settled, perpetually focused on the gap between where they are and where they need to be.
"The thing about being a founder CEO is no one is there to rescue you. Your parents aren't gonna rescue you. Your VC is not gonna rescue you."
Brian HalliganThe double diamond — that sacred diverge-converge-diverge-converge process that design schools taught as gospel — is dead. Engineering with AI moves so fast the old cycle simply cannot keep up. Mocking was 60–70% of the design job. Now it's 30–40%, and shrinking fast.
What replaces it? Two modes. Mode 1: Real-time implementation — designers in the IDE alongside engineers, shipping the same day. Mode 2: Short-horizon vision — not 5-year roadmap decks but 3–6 month prototypes that actually run and can be tested immediately.
But amid all the speed, Jenny insists on one non-negotiable: someone still has to own the decision. AI doesn't remove the need for judgment — it amplifies the cost of bad judgment. Her hiring archetypes for the new era: "block-shaped" generalists, deep craft specialists, and cracked new grads unburdened by old rituals.
"It's not just designers feeling like we have to keep up with engineers. Even engineers are like — how do we keep up with ourselves? How do we keep up with our agents?"
Jenny WenAn investor told Grant Lee his idea was the worst he'd ever heard — then hung up. Grant was pitching from a kitchenette in London at 11 PM, kids asleep down the hall. Five years later, Gamma has $100M+ in annual revenue, a $2.1 billion valuation, and 50 million users — served by 30 people.
Their Product Hunt launch spiked and flatlined. Real product-market fit came from obsessing over the first 30 seconds of the experience — making onboarding feel magical. Organic word-of-mouth now drives more than half of all new subscribers.
The growth engine that actually scaled: thousands of micro-influencers in niche communities — educators, small business owners, regional creator circles. Not mega-influencers. Grant personally jumped on a call with every early creator. "You want them to tell your story in their voice." The result was wildfire.
The operating tempo is relentless. Morning idea. Afternoon prototype. Evening data. Ship by end of day, every day. The team is built entirely on generalists and player-coaches — nobody who only does one thing. At 30 people for 50 million users, that's not lean. It's a statement about what's possible now.
"The investor pauses and says — that has to be the worst pitch, worst idea I have ever heard. You are never going to succeed. And before I could even respond, he hangs up."
Grant LeeDifferent people, different domains. The same patterns kept appearing:
Boris runs 5 agents in parallel. Grant ships by EOD. Jenny's vision horizon shrank to 3 months. Brian sees a Cambrian explosion. Velocity is the new moat.
Boris says "builder" not "engineer." Jenny wants block-shaped designers. Grant runs 30 generalists for 50M users. Narrow specialization is a liability.
Jenny says accountability is non-negotiable. Brian insists on DRIs. Grant personally onboarded every influencer. Speed without ownership is just chaos.
The double diamond. The "engineer" title. The mega-influencer playbook. The 5-year roadmap. Every guest killed one — and offered something sharper in its place.