Execution

Unshipped work has zero value—relentless focus on completion, not progress

Design systems that force you to stay on track toward your goals

Three-stage progression from fluency to process optimization to growth inflection

Three brain wave states that enable different types of thinking—and why we spend too much time in beta

Black-and-white rules are easier to keep than conditional ones

Identify the biggest bottleneck, solve it deeply, pick the next

Ultra-low-fidelity design artifacts that force discussion of concepts, not aesthetics

Don't choose between building or buying—do both strategically

Rebuild trust during crisis through proactive, frequent communication with explicit commitments

Balance your portfolio between big fundamental bets and small incremental experiments

Three groups emerge during organizational transformation—strategy must differ for each

At scale, shift from controlling decisions to setting culture and principles that guide distributed teams

Product management distilled to two core skills: bringing clarity and having conviction

Code quality and product success are uncorrelated—ship value, not elegance

Assign teams to deeply understand competitors, then present findings in a competitive summit

For every unit of work, make the next unit of work easier to do

An iterative AI product lifecycle that builds trust through gradual autonomy increases

Annual company-wide solicitation of high-risk, high-reward ideas

The biggest RAG gains come from better data preparation, not better vector databases

Prioritize discovery on your biggest swings to enable true innovation

Actively delete code to simplify codebases and move faster

People who can design, build, and ship complete prototypes—enabling innovation at scale

Use visionary prototypes and Loom walkthroughs to keep teams aligned and motivated

Present with recommendations, not information dumps—leaders can only hold three facts at once

Only discussion belongs in meetings—discover and decide independently

Start with manual processes, identify what works, then build technology to automate at scale

Run discovery and delivery simultaneously to de-risk assumptions while shipping

Invest in evals proportional to scale, risk, and competitive importance

If the model is the product, the eval is the product requirement document

Treat customer support as product feedback by having support report to product

Bucket work into three types to set appropriate expectations and investment levels

Public commitments to others create the forcing function that self-discipline cannot

Design experiments to maximize treatment effect so failures are definitive learnings, not inconclusive results

Simplicity is worth fighting for—and it requires fighting for

Distinguish between improvements that prevent bad outcomes vs. improvements that enable great outcomes

Weekly ritual for recognizing wins across the company to build team morale

Goals, Audience, Creative, Channels, Stakeholders—a marketing brief that prevents wasted work

Choose between developer-led, PLG, or direct sales based on buyer/user alignment

Don't stop until you've reached first-principles truth—expert answers are often incomplete

Optimize help centers to capture long-tail questions in LLM answers

The PM job is to help teams ship the right thing at the right time in the right way

Derive key results from objectives by asking 'How do we know we succeeded?'

A simple format for manager updates that covers the work, the team, and yourself

A team ceremony that prioritizes learnings over status updates to drive continuous improvement

Three-part narrative structure for strategic planning that forces clarity

Create alignment at scale by making prototypes spread virally within the company

Planning process that alternates between teams and leadership for iterative synthesis

Maximize output per unit of time instead of extending hours

The final 5% of effort is what separates good from excellent

Drive AI adoption by having executives personally use the tools daily, not by mandating adoption

During once-in-a-lifetime opportunities, execute with zero regret intensity

Concentrate marketing in intense bursts rather than spreading evenly across time

Obsess over the feedback loop cycle, not your functional specialization

Keep burn rate extremely low to maximize the number of strategic bets you can take before finding product-market fit

Every engineer ships a feature that could convince someone to pay—every single week

Annual deletion of all recurring meetings to reset calendars and reclaim crafter time

Set the floor for process, but explicitly encourage going higher

A lightweight process for launching new products: validate fast, polish before scaling, then leverage distribution

Fix every bug immediately—no backlog, no prioritization

Map how your product spreads through organizations by visualizing user invite networks

Set pace expectations by bringing every timeline in one iteration

Start creative projects with a limited commitment to test whether you love the work before going all-in

Create alignment by having senior leaders maintain one continuous conversation with a single roadmap

A PM-to-PM forum for debating and validating problem spaces before solutioning

Operations roles should aim to eliminate themselves through automation and process—not to empire build

Price your product based on measurable outcomes delivered, not usage or seats

Sustainable success is built on years of foundation work that becomes invisible once you break through

Two people on one machine for higher quality output, faster learning, and intense focus

Frictionless problem reporting with a single button click to capture developer pain points

Replace top-down OKR approval with fast peer review from three connected teams

Roll out new processes by starting with your highest-performing team first

Internal platforms need product management—treat them like products, not just infrastructure

The first design artifact sets the baseline—delay drawing as long as possible to preserve creative options

Meetings are a combination of priming and decision-making—skipping priming is where most meetings fail

Start with the problem, not the technology—especially when the technology is exciting

Process brings everyone to the average—including your best performers

Commit to dates only for work in front of you through four distinct phases

Filter ideas ruthlessly—lots at the top, few at the bottom—don't ship everything that enters

Create systems that allow product teams to thrive, not just hire people

Track the context of user actions (properties), not just that actions occurred (events)

Measure success by whether stakeholder questions shift from 'how' to 'what next'

For any new launch, you can optimize for two of three constraints—quality, features, or deadline

A quarterly goal-setting system that creates focus, alignment, cadence, and learning

People can know something is coming but still not be ready—bridge the gap

Major system rewrites are usually a trap—evolve incrementally instead

Train AI through simulated real-world environments where models learn by pursuing rewards across multi-step tasks

Find what's working and do more of that

When time pressure hits, cut scope ruthlessly—never quality

Plan around secular industry shifts, not fixed time periods

Run new ventures with completely separate teams, space, cadence, and culture from the parent company

Visualize complete user journeys with real personas and exact situations

Give teams one goal, one thread, and shield them from everything else

Commit fully for six months to validate a new venture before deciding to continue or quit

Your output is constrained by your slowest bottleneck—accelerating one part without addressing others creates new problems

Software built for exactly one person—yourself—with no obligation to maintain it for others

Don't boil the ocean to make a cup of tea—launch tiny experiments and build from what works

Taking on technical debt is how startups operate faster than larger companies—use future employees' time now

A process for becoming inventive through dedicated time, expertise, and combining existing concepts

The art and sophistication of deciding what kind of AI model to create shapes its capabilities more than raw compute

Build trust with AI agents the same way you would onboard a new human teammate

Find where technology creates the most value in your business, then ruthlessly focus there

Allocate more compute to inference for better results without changing the base model

Ask 'what would you do with 20% more time or double the team?' to surface ambitious ideas

Build proprietary AI where you have advantage, partner for commodities, enable ecosystem innovation

Ensure meeting outcomes stick with three closing questions about decisions, actions, and communication

Tools are just meant to solve problems—not the other way around

Evaluate the path to a goal, not just whether the goal was reached—how you get there matters

Product and operations as two engines of a jet plane—both must work together for efficiency

De-risk execution by investing in understand work before building

Prioritize shipping speed as the core organizational value

Review product flows together as an entire company to build shared language and standards

Monday commitments and Friday celebrations create rhythm for sustained progress

Focus on users, data, and prompts—not chasing the latest AI news

Fake the product experience with manual processes to validate ideas before building

Start with the customer problem, write a press release, then work backwards to build it

Maintain perfect information transfer during hypergrowth through daily all-hands synchronization

Build new products within large companies by zigging when others zag and avoiding the limelight