← Articles

All published articles

84 articles in one lightweight list, newest first.

  1. What I Expect From a Product Manager Now

    Nov 12, 2026

    I expect PMs to help build the first version, work through technical questions with specialists, and use more of their time to learn from customers.

  2. Pac-Man, in Real Time

    Nov 9, 2026

    The Pac-Man algorithm I developed reached the maximum score in our game under its four-life settings: 3,333,180 points.

  3. While My Colleagues Tackle Millennium Prize Problems, I’m Tackling Pac-Man

    Nov 5, 2026

    Building a browser game and a verified replay to prepare for a real-time Pac-Man algorithm.

  4. We Need a New Operating System

    Nov 3, 2026

    That calls for a new OS designed for people and agents to work together, with responsibility for carrying intention through to a result across applications and devices.

  5. Drawing the Logic Behind a Complex Decision

    Nov 3, 2026

    Narrated drawings offer a way to inspect complex systems by revealing relationships in sequence and letting people reason through a changed case.

  6. Article

    What Tetris Taught Me About AI Decisions

    Oct 29, 2026

    A Tetris experiment explores when a model earns its place alongside a conventional algorithm, and what that means for recommendations and product decisions.

  7. Article

    How I think about evals for content discovery

    Oct 27, 2026

    Evals make a discovery team's expectations testable, connecting individual failures, ranking comparisons, and user evidence to decisions about what deserves to ship.

  8. Article

    Your Dreams Are My Memories

    Oct 26, 2026

    Experience can sharpen an ambitious product proposal when we can explain what earlier attempts actually established and what has changed.

  9. Article

    We Want Your Business. Please Come Alone.

    Oct 23, 2026

    Customers who delegate to AI assistants still deserve service, with safeguards for authority, automation costs, abuse, and the businesses receiving their requests.

  10. Article

    My AI Product Manager Needed a Process

    Oct 22, 2026

    The AI product manager experiment exposed a coordination problem. Infinite Queue and an orchestrator give our AI agents defined roles and a workflow to follow.

  11. Article

    Orchestrating the Software Lifecycle

    Oct 21, 2026

    Building an AI system to coordinate the software lifecycle, with human judgment at key decisions.

  12. Article

    I Hired an AI Product Manager. Then I Had to Take the Project Back.

    Oct 19, 2026

    The Aviary Frog shipped; the parallel work that followed grew hard to reconcile, and I took the project back.

  13. Article

    What Would You Trust a Streaming Agent to Do?

    Oct 17, 2026

    A streaming agent earns the invitation to choose by interpreting the current request, retaining useful context, and making an imperfect choice easy to redirect.

  14. Article

    Following a Metric Into the Product

    Oct 15, 2026

    AI agents can carry a product investigation across data, code, and context while keeping the evidence open to inspection and correction.

  15. Article

    A Successful Launch Can Still Be a Product Failure

    Oct 13, 2026

    A successful release can leave the original customer problem unresolved. Post-launch investment needs evidence from the complete task and people able to act on it.

  16. Article

    Learning to Manage Bob

    Oct 10, 2026

    Delegating the toucan project to an AI product manager exposed unclear expectations for creativity, independence and measurable success.

  17. Article

    Discovery Does Not End at Handoff

    Oct 8, 2026

    A working implementation can overturn a reasonable design assumption. The difficult part is letting that evidence change the right decision without reopening the entire product.

  18. Article

    What Makes a Great Product Team

    Oct 6, 2026

    Great product teams combine specialist depth with shared responsibility for customer usefulness, from the first question through delivery and learning in use.

  19. Article

    How Much Should One Person Be Trusted to Direct?

    Oct 1, 2026

    People with high agency and taste could direct far more work through AI agents. Their budgets, scope, and compensation should reflect the impact they create.

  20. Article

    What does a builder team look like?

    Sep 29, 2026

    A small team can use AI to experiment across the product, backend data, and algorithms, shortening the wait between an idea and what it teaches us.

  21. Article

    What Would It Take to Build Together?

    Sep 23, 2026

    A small team can explore a product idea in working code when its members have shared time and room to change existing commitments.

  22. Article

    The Stakeholder Is Not the Customer

    Sep 17, 2026

    Stakeholders may also be customers or hold better customer evidence. Product judgment translates the organizational fact inside a requested solution and names its authority.

  23. Article

    The Feature You Kill Defines the Strategy

    Sep 15, 2026

    Strategy becomes credible when a team records the attractive work it declined, the customer promise it protected, and the evidence that would reopen the choice.

  24. Article

    Roadmaps Should Have Expiration Dates

    Sep 10, 2026

    Roadmaps should coordinate today's choices without granting permanent authority to the assumptions that produced them.

  25. Article

    Good Judgment Starts Before the Dashboard

    Sep 8, 2026

    Good product judgment does not outrank data; it matches the next move to the available evidence and defines what could prove the judgment wrong.

  26. Article

    The Decision Is the Deliverable

    Sep 3, 2026

    A precise decision question tells a product team which evidence matters, how much confidence the choice requires, and who must own it.

  27. Article

    The Student Still Owns the Thinking

    Sep 1, 2026

    An AI tutor can keep office hours open at any hour. The student still has to understand and account for the work.

  28. Article

    What College Is For When Every Student Has an AI Tutor

    Aug 27, 2026

    When broad instructional help is always available, a university's scarce value becomes the people who challenge students and the work they learn to do together.

  29. Article

    What Did the Demo Prove?

    Aug 20, 2026

    Demos, dashboards, and experiments can all be honest while still creating false confidence. Product value depends on matching the evidence to the claim and the next decision.

  30. Article

    How I Build Presentations with Codex and Other Frontier Tools

    Aug 18, 2026

    A generated deck should be designed for the next human edit, from the number of objects on a slide to the template, notes, and companion versions around it.

  31. Article

    Where the Smartest Model Belongs

    Aug 13, 2026

    Model routing should follow the cost of being wrong: concentrate frontier reasoning where errors spread, then bound the remaining work with explicit contracts and evidence.

  32. Article

    When Source Code Outruns Human Review

    Aug 4, 2026

    As AI-generated code outgrows human attention, trust will shift from routine source inspection to evidence, controls, and risk-based escalation.

  33. Article

    When the Idea Answers Back

    Jul 30, 2026

    Language compresses a product idea. AI gives high-agency product people more power to build enough of it to explore, learn from its behavior, and discover what the original words left out.

  34. Article

    The A/B Test That Lies

    Jul 28, 2026

    A/B tests can establish causal movement, but teams still need a metric hierarchy that distinguishes local engagement from durable product trust.

  35. Article

    Your Team Needs Fewer Dashboards and More Dials

    Jul 23, 2026

    Metrics become management only when teams connect signals to authorized controls, decision rights, guardrails, and a memory of what happened next.

  36. Article

    Your AI Coworker Should Know Which Doors It Can Open

    Jul 21, 2026

    A product manager should understand the business consequence of an action. They should not have to become a systems administrator before asking an AI tool to help with their work.

  37. Article

    AI Agents Need Breakpoints

    Jul 18, 2026

    Codex can order queued turns, but it cannot yet express a durable approval gate between them or constrain the current turn to the phase the user intended.

  38. Article

    What Should an AI Be Allowed to Know?

    Jul 18, 2026

    An AI AMA earns trust by separating general expertise from personal claims, screening every runtime surface, and treating privacy failures as release blockers.

  39. Article

    The Art of Making Algorithms Understandable

    Jul 14, 2026

    AI product managers need faithful explanatory artifacts that turn complex recommendation systems into reviewable behavior, not just static requirements.

  40. Article

    One Task, Multiple Agents, One Continuous Loop

    Jul 13, 2026

    Trustworthy multi-agent systems emerge from evidence contracts, interruption rights, recovery paths, and human feedback—not from parallelism alone.

  41. Article

    I Knew What AI Could Do. I Forgot to Think That Way.

    Jul 11, 2026

    AI-forward thinking begins by questioning inherited architecture: let models interpret raw evidence, deterministic systems constrain and test it, and humans own the decision boundary.

  42. Article

    The Best AI Will Not Ask You to Choose a Model

    Jul 10, 2026

    AI products will earn broader adoption when they translate user intent into the right capability, instead of making people configure the machinery.

  43. Article

    Don’t Inherit the Tool. Inherit the System.

    Jul 8, 2026

    AI-assisted work becomes portable when teams preserve runnable systems, clear decisions, tests, and documentation beyond any one tool.

  44. Article

    The Discovery Brain: Where LLMs Can Strengthen Streaming Recommendations

    Jul 7, 2026

    LLM-native discovery can unify streaming recommendations when grounded catalog truth, governed memory, validation, and feedback protect user trust.

  45. Article

    The Camera Is Becoming Software

    Jul 6, 2026

    Software is turning cameras into adaptive capture systems, pushing dedicated gear toward high-intent work while phones own everyday memory.

  46. Article

    Fit for Purpose: Choosing the Right Quality Bar

    Jul 3, 2026

    Teams move faster when precision serves the user experience, not internal certainty, and good enough means fit for purpose.

  47. Article

    Modernizing Discovery and Promotion in Personalized Products

    Jul 1, 2026

    Modern discovery systems balance personal relevance, popularity, momentum, and strategic intent without letting promotion overwhelm user value.

  48. Article

    Why Strong PMs Are Becoming Builders

    Jun 29, 2026

    AI is drawing product managers closer to systems, prototypes, evaluation, and the cross-functional judgment required to shape them well.

  49. Article

    The Builder Team: How AI Changes the Shape of Product Development

    Jun 25, 2026

    AI-native product teams can reduce handoffs through builder pods with stronger shared context, executable specs, earlier evaluation, and clear human ownership.

  50. Article

    The Modern Wildlife Photography Mindset: Protect the Shutter Speed, Trust the ISO

    Jun 22, 2026

    A sharp, slightly noisy wildlife image usually beats a clean miss; modern cameras make that exposure tradeoff easier to trust.

  51. Article

    The Evaluation Layer Is Becoming the Product

    Jun 22, 2026

    AI product teams need evaluation as a visible product surface, not a late report card, because fluent demos can hide weak or unsafe judgment.

  52. Article

    The Work Is Not Done Until It Is Reviewable

    Jun 21, 2026

    AI-assisted work is only mature when the artifact carries a review surface: intent, evidence, checks, privacy, and human judgment.

  53. Article

    The Product Manager’s New Operating System

    Jun 8, 2026

    AI-native product work rewards PMs who can turn intent into artifacts, inspect evidence, and own judgment across faster operating loops.

  54. Article

    Building a Photographer’s Guide with LLMs

    Jun 7, 2026

    A Bosque photography guide becomes a product case study in turning messy research into trusted, structured, useful AI-assisted artifacts.

  55. Article

    The AI Adoption Gap Is Becoming a Planning Problem

    Jun 5, 2026

    AI-assisted teams and traditional planning rituals are now using different clocks, forcing organizations to recalibrate estimates, expectations, and execution habits.

  56. Article

    We Risk Building the Wrong Things Faster

    May 27, 2026

    AI can help companies build faster, but without sharper product judgment, stronger architecture, and outcome-based measurement, they risk creating more software without more user value.

  57. Article

    AI Homework

    May 26, 2026

    AI adoption becomes real when teams rebuild one recurring workflow around sharper intent, visible evidence, verification, and human-owned judgment.

  58. Article

    Human Intent Is the Next Operating System

    May 25, 2026

    Frontier tools can become a new operating layer between human intent and digital services, reducing the cognitive load software has pushed onto users.

  59. Article

    The Next-Week AI Playbook

    May 25, 2026

    Teams can make AI adoption concrete by redesigning one real workflow around evidence, reviewable artifacts, and accountable human judgment.

  60. Article

    Why Conversation Will Become Part of Streaming Search

    May 24, 2026

    Streaming search is moving from exact-title retrieval toward conversational discovery that understands mood, context, constraints, and catalog truth.

  61. Article

    When AI Produces the Artifact, Measure the Work Beneath It

    May 23, 2026

    AI makes visible work artifacts cheaper, which raises the standard for judgment, accountability, and the real work underneath the work.

  62. Article

    The Frontier Recommendations PM

    May 22, 2026

    Frontier coding agents can help product managers connect user intent to algorithmic system behavior across recommendations, search, personalization, and AI discovery.

  63. Article

    Do More With the Same, Not the Same With Less

    May 21, 2026

    AI should help teams do more with the same people by reducing handoffs, increasing builder range, and giving product judgment more cycles.

  64. Article

    The 20/70/10 Rule for Working With AI

    May 20, 2026

    The 20/70/10 model keeps AI workflows fast without letting speed replace context, verification, judgment, or ownership.

  65. Article

    Why I Built Codex Log Viewer: Treating User Intent as a First-Class Audit Trail

    May 19, 2026

    Codex Log Viewer turns local Codex sessions into a searchable, privacy-conscious audit trail for the human intent behind AI-assisted software work.

  66. Article

    Using AI to Make Promotion Review Preparation More Consistent

    May 13, 2026

    An experimental workflow for making promotion review preparation more consistent and evidence-backed while preserving privacy, governance, and human accountability.

  67. Article

    What Is AI? How Does It Impact Us?

    May 7, 2026

    AI is a new capability layer that can make product teams faster, more creative, and more effective, but only when paired with strategy, judgment, and responsible guardrails.

  68. Article

    How Much More Can We Do With the Same?

    May 6, 2026

    AI changes how product teams ideate, prototype, evaluate, and launch. The real leadership challenge is turning that speed into better judgment and customer value.

  69. Article

    The Missing Layer in AI Adoption

    May 1, 2026

    AI changes who can contribute. But only if companies remove the friction between having an idea and making something real.

  70. Article

    Move Fast, But Don’t Break the Magic

    Apr 28, 2026

    Why AI product teams need release systems that protect quality, trust, and the workflows users depend on.

  71. Article

    How AI Is Moving Team Collaboration Into Working Software

    Apr 26, 2026

    AI-assisted development is making prototypes a shared workspace where teams can shape product ideas, narrative, and experience inside working software.

  72. Article

    Taste, Trust, and Truth: How to Evaluate AI-Powered Discovery

    Apr 21, 2026

    Once a discovery system starts speaking, accuracy alone is not enough. You have to evaluate fit, trustworthiness, and product impact together.

  73. Article

    Five Advantages That Matter More as AI Expands

    Apr 19, 2026

    As AI expands the supply of plausible outputs, taste, context, exclusivity, trust, and distribution become more important sources of durable value.

  74. Article

    How I Would Train a New PM for an AI-Native Discovery Team

    Apr 14, 2026

    A ninety-day approach to product management in search, personalization, experimentation, and learning for AI-native discovery teams.

  75. Article

    The Early Adopter Tax

    Apr 13, 2026

    The real return from early AI adoption is not mastering temporary tactics. It is building the judgment to evaluate, redesign, and govern the work before the tools make it look easy.

  76. Article

    What Early-Talent Recruiting Reveals About Leadership

    Apr 3, 2026

    Early-talent recruiting makes a company’s leadership standards visible through communication, follow-through, and respect for candidates’ time.

  77. Article

    Programmer vs. Computer Scientist

    Mar 28, 2026

    Why AI getting better at writing code does not make computer science less valuable; it makes depth of judgment more important.

  78. Article

    How LLM Coding Could Change Source Control

    Mar 24, 2026

    LLM coding tools could make version control less visible by capturing meaningful states, intent, and recovery paths as part of the work.

  79. Article

    Lessons from Building with AI: Developing the Connect-K Game

    Mar 17, 2026

    From a 3D cube in space to a published multiplayer experience — key lessons on working with AI coding tools that every builder should know.

  80. Article

    How AI Tools Are Raising the Bar for PMs

    Mar 11, 2026

    What building software with my son taught me about the future of product management when the gap between idea and working software collapses.

  81. Article

    How New Graduates Can Build Leverage in an AI World

    Mar 9, 2026

    Entry-level product work is changing, giving aspiring PMs new ways to demonstrate judgment, technical range, and end-to-end ownership.

  82. Article

    Lean In, Learn, Build: How Did You Become a Better Product Manager This Week?

    Feb 14, 2026

    Frontier AI is changing product work quickly, making continuous learning, practical building, and durable curiosity essential habits for PMs.

  83. Article

    Python or Lovable? The Role of Modern AI Tooling in Product Management

    Feb 8, 2026

    Should aspiring Product Managers invest in learning Python, or double down on modern AI tools like Lovable, v0, and Replit?

  84. Article

    How LLMs Can Improve the Search Experience: A Shopping Example

    May 4, 2025

    A bike helmet shopping journey shows where conversational search improves a decision and where missing constraints still weaken the result.