Think Like a Gardener,
Not a Builder

Alex Komoroske

Ex-Google (Chrome, Maps) · Ex-Stripe Head of Corporate Strategy
Founder reimagining the web for the AI era

LENNY'S PODCAST

Core Framework

Gardener vs. Builder Mindset

"If you do this properly, it looks like magic. I've been told this is completely against all the advice people get, but I think it's a very powerful approach."

Builder mindset: have a plan, manipulate reality to match it
Gardener mindset: find things that can grow on their own, then direct and curate
Builders are bounded by effort; gardeners can unlock compounding returns
Most valuable org, product, and career moves share the gardener pattern

Deep Dive

LLMs as Magical Duct Tape, and Why Taste Wins

17K+

unpublished working notes in The Compendium

600pg

Bits & Bobs Google Doc, published weekly

13yr

at Google: Search, Chrome, Maps, AR

LLMs = magical duct tape: distilled societal intuition at a cost structure between human and plain computing
Old assumption: software is cheap to run, expensive to write. LLMs undermine both
The "5% punches you in the face" problem: LLMs are squishy, design for failure, not just success
Don't use LLMs as oracles; ask instead what new things can you now build with magical duct tape
Taste is the scarce resource: when production cost collapses, the differentiator is your distinctive perspective

The taste test Differentiate from what the LLM would've written given the same prompt. How distinctive is your perspective from the average? That gap is your competitive moat.

Playbook

Slime Mold Orgs & the Weekly Reflection System

The Slime Mold Insight

Coordination cost grows with the square of the number of people. As you scale, your company stops being a sports car and becomes a big rig, driving it like a sports car causes crashes.

Two valid choices: one coherent big rig (Apple) or a swarm of autonomous sports cars (AWS). Fighting the dynamic is the fatal mistake.

Metaphor as strategy weapon

Connect 9 of 10 dots and let the listener connect the last one. This lets you say controversial things safely, people self-apply the insight and feel ownership of it. The more diverse the audience who finds it resonant, the wider the idea can spread.

Alex's weekly reflection loop

Mon–Thu: back-to-back meetings; capture notes instantly on phone
Every 1–2 days: process into Compendium working notes, fix typos, add context
Friday afternoon: flag notes that still resonate; export to Google Doc
Weekend (kids' nap): distill into long-form Bits & Bobs insights
Monday morning: publish publicly, for himself first, audience second

"The mundane, pointless bullshit will take every square inch you give it. You got to make the space to sit back and reflect and luxuriate in these ideas."

Tactics

Nerd Clubs, Salons & Finding Superpowers

Strategy salons / nerd clubs: Alex's "secret weapon", small, recurring, cross-disciplinary gatherings to test ideas with diverse audiences
A resonant idea in sales AND engineering has a much larger implied ceiling than one that only lands with one group
Each time an idea resonates with a new audience, invest a bit more time tightening the framing
Superpower identification: within the first session or two of mentoring, articulate "I think your superpower is…", it unlocks outsized effort and receptivity to feedback
Treat everyone as the Buddha: find seeds of greatness in everyone, even if they don't see it themselves

The adjacent possible In the community gardening phase of a technology (not factory farming), curiosity and play beat execution velocity. Explore weird possibilities like WebSim, things that could only exist in an LLM world.

Contrarian

Alex's Most Counterintuitive Takes

✗ More execution = more value created INSTEAD → ✓ Builder effort is capped by input. Gardener-style work compounds, you direct growth rather than manufacturing output, and it can far exceed your effort.

✗ LLMs will mostly replace the need for taste and judgment INSTEAD → ✓ When production cost collapses, the world floods with slop. Taste, your distinctive perspective that diverges from the LLM average, becomes the only durable competitive edge.

✗ Big organizations should fight coordination friction with more process INSTEAD → ✓ Coordination cost grows with the square of headcount, it's a law of nature, not a fixable bug. Embrace your org as a slime mold: autonomous swarms find solutions you never knew you were searching for.

✗ Reflection time is a luxury you earn after you're productive enough INSTEAD → ✓ The Friday with no meetings is what makes the other four days multiply. Deep thinking is a force-multiplier, not a reward, and if you don't schedule it, the mundane will consume 100% of your calendar.