Field Notes

The only PM blog written by someone who still ships code every week. Failed experiments included. Agent recipes you can fork.

2026

Absorb Pain, Excrete Product: The FDE HandbackPalantir made Fujitsu a Global FDE Partner. When someone else does the deploying, the handback has to be a written packet with a date, and a decision back.
Kill the System-of-Record SlideServiceTitan cut off Podium, kept 1,000 customers, and added zero revenue. Why 'we are the system of record' is a retention fact, not a growth slide.
2,000 of 17,000: The FDE Shortage Is a Product ProblemOnly 2,000 of 17,000 US forward deployed engineers can deliver real ROI, and demand is up 2,100%. The gap is not talent. It is what the product lets an FDE do.
What 267 Posts Are For: A Map of the Field Log267 posts, 49 handbook chapters, 203 answers, 115 templates. What each part is for, who it was written for, and where to start from your seat.
Nobody Is Reading AnythingThe viral voxium post is not about Claude Code. It is about a company that made generation free, never funded the reader, and the fix for that.
Percent Migrated Is a Vanity MetricWe modeled 50 pricing disputes a week and got 600. Percent migrated said the migration was done. The exception rate said nobody believed it yet.
I Automated My Reading. The Filter Was Wrong Three Times.A filter over 45 sources cut 102 posts to 34. Version one dropped the best piece of the week. Here is each bug, with the titles that exposed it.
Every Check Was Green. Four Sources Were Dead.My source monitor reported 43 of 45 feeds healthy. Four had published nothing for 72 to 612 days. Every one returned HTTP 200 and parsed perfectly.
Which Door Keeps the ReceiptFour interfaces writing to one record only stays one record if all four write back. Whichever keeps the richest local trace becomes the real system of record.
The AI Eval Starter Kit: Six Templates, Error Analysis to GateTeresa Torres's evals method as six copyable templates: error analysis log, type chooser, code assertions, golden dataset, judge prompt, and the gate.
Torres Wrote the Evals Guide PMs Needed. Here Is Step Four.Torres's three-step evals guide is the best hands-on primer for product teams. I'd add a fourth step, a cost column, and one warning about the cheap tier.
Benioff Said AI Replaces the UI. Then He Named the Product.At Dreamforce 26 Benioff said AI replaces the UI and Stokes said the value is trust to hold your data. Read together, that is Salesforce's new product.
The FDE Is the Product Builder Wearing a Visitor BadgeOpenAI raised $4 billion for a deployment company. Anthropic launched a $1.5 billion one. The role they hire for is the Product Builder, from the services side.
The Enforceable HalfSpotify cut agent token costs 90% by routing grunt work to a cheap model. The useful finding is that they could only enforce half their own rule, and why.
Cagan's Definition Is Right. The Weights Are Wrong.Cagan maps Benedict Evans's three product skills onto his risk framework. The skills are right. Their AI exposure is wildly different, and that is the story.
Salesforce Named the Harness. The Missing Piece Isn't on the Slide.Salesforce named the enterprise AI harness: six capabilities, a control plane, a 2027 window. The layer that decides whether it works cannot be bought later.
Context Is King. Written Context Is Rented.Everyone is racing for the context layer. The quadrant nobody occupies is empty for a structural reason, and importing your wiki will not fill it.
The Canvas Was Never the WorkMiro sold at 2.3x ARR while profitable with $435M in net cash. Compression explains the range, not the position. The mechanism is legibility.
Polish Stopped Being EvidenceFidelity used to be a proxy for conviction in design review. AI broke that proxy, so the work that looks most finished now gets the shallowest critique.
Everyone Wants to Be the Layer UnderneathGlean and Alation arrived at the same posture from opposite ends: be the substrate, not the assistant. You cannot own a daily habit and stay neutral.
Salesforce Put Its CRM Inside ClaudeSalesforce beat on GAAP because it owns a piece of Anthropic, then shipped its CRM inside Claude. The app layer just volunteered to be the substrate.
The Wince Is the Spec: Bottom-Up Evals a Model Can't WriteClaude writes half your evals. It's the half that catches nothing. Top-down vs bottom-up evals, and how I turn gut reactions into verifiers on Heidi.
Cost Gets You In. Usage Keeps You.Glean's jump from $100M to $300M ARR got read as growth. The better read: why a buyer approves your product and why they keep using it have come apart.
SpaceX Just Bought Cursor for $60B. What It Actually Bought.SpaceX closed its $60B all-stock buy of Cursor on Aug 14. Not a model story, an operating one: the choices that made a 2022 MIT project worth $60 billion.
The Harvey Playbook: How AI Cracked Big LawHarvey hit $350M ARR and an $11B valuation selling to Big Law. The six operating choices behind it, and the one place the playbook could break.
The IKEA Story Does Not Prove What You Think It ProvesSalesforce fired 4,000, IKEA reskilled 8,500, and LinkedIn decided this settles the AI question. The facts are real. The comparison is junk.
100 to 3,000 in a Week: Why Microsoft's Best Agent Is Not Called CopilotMicrosoft's Scout agent, formerly ClawPilot, went from 100 to 3,000 users in a week with no mandate. What its naming, identity, and org design teach.
Every Artifact Is Somebody's Promotion PacketThe PM ladder runs on documents because documents are the evidence promotions are built from. When AI collapses the artifact, the currency disappears.
Malleable Software Was Never the Point. The Loop Is.Dave Killeen's Dex files its own bug reports: your agent tells his agent what broke, the fix ships. Why malleable loops beat malleable software.
One Org That Got Landing RightSnowflake made landing its business model: it only gets paid when customers use the product. How consumption pricing solved what an org chart could not.
The Eval That Caught What the Demo Did NotThe demo was flawless. The eval I ran against 200 real cases failed on a slice that was a fifth of our traffic. The failure the demo structurally hid.
The Landing Ledger: What It Costs to Build vs What It Costs to LandBuilding costs collapsed. Landing costs did not. Line by line, where the money and attention actually go, and why orgs starve the cost that dominates.
The Launch That Worked and Did Not LandEvery launch metric was green: press, signups, week-one activation. Two months later the usage curve was flat. The exact place it fell through.
The Person Whose Job This Shift ThreatenedThe Product Builder shift is a thesis until it has a face. The best PM on my team, whose craft was the thing AI made cheap, and how I handled it badly.
The Repricing, Beyond AirtableAirtable at 2.7x ARR is not an anecdote. Chegg, Stack Overflow, Squarespace, and Smartsheet are the same repricing, in two kinds. Which one is yours?
The Sierra Playbook: What They Actually Do DifferentlySierra hit $100M ARR in seven quarters and $200M by mid-2026. Not a model story: six operating choices on pricing, accountability, and go-to-market.
The Week the Prototype Beat the SpecA twelve-page spec we argued over for three weeks lost to a prototype I built in an afternoon. The number that ended it, and what I stopped doing.
What Landing Actually IsShipping is not landing. Landing is when a customer's behavior durably changes and the value sticks. A working definition, and the four ways it fails.
What Replaces the PM Ladder When the Deliverables Are GoneIf documents were the evidence promotions ran on, and AI collapsed them, leveling needs a new currency. What counts as evidence from L4 to Principal.
31,832 PMs Applied. Whatnot Hired One. The Premise Is the Point.Whatnot runs on the premise that it regrets product management existing: 22 senior PMs, $8B GMV, managers doing 90% IC work. Where it corrected me.
Who Owns Landing? Right Now, NobodyLanding falls in the gap between product, marketing, sales, and success. Each hands off before behavior changes. Everyone's concern, and nobody's job.
Airtable Sold for 2.7x ARR. That Number Is the Whole Lesson.Airtable did not fail. It has ~$480M ARR growing over 20%. What collapsed was the story investors were paying for, and the AI pivot could not save it.
Twelve Builders, Twelve Products, Zero LandingsThe 2026 CPO Insights Report says product builders replace PMs by 2030. The role is real. The framing points at building, just as building stopped binding.
What Anthropic's Dianne Penn Taught Me About the FrontierWho Dianne Penn is, what she leads at Anthropic, and the four ideas she argues for: evals as the new PRD, the jagged edge, product overhang, and judgment.
How I Build Eval RubricsThe eval rubric method I actually use: grade thirty outputs first, tie every dimension to downside exposure, and go binary wherever you can.
The PM Does Not Make Decisions. Make Sure They Get Made.The PM's job was never to make decisions, it was to make sure decisions get made. In the AI era, when agents build in a day, that job becomes the whole game.
AI Can't Understand the User. That User Is Disappearing.The best defense of the human PM is that AI can't understand the user. But that user is a departmental role, and agents are dissolving the departments.
The Agent Layer Is Not the Catalog. Enterprises Will Learn This the Hard Way.Alation's AIOS rebrand says what most agent builders miss: it is not trying to be the agent platform, it is trying to be what every agent plugs into.
Writing for Machines Is the New Writing for ExecutivesHalf your readers are now agents. Writing that executes faithfully is its own craft: specs as eval sets, prompt hygiene, GEO basics, and the cold-read test.
Is This Product Bet Group Stage or Knockout? Reversible vs. IrreversibleThe World Cup is two tournaments bolted together. Knowing whether a product decision is group stage or knockout is the same question as downside exposure.
Stage Wins or the Yellow Jersey: You Don't Win the Tour by Winning StagesThe rider with the most stage wins almost never wins the Tour. A model for AI product management: cumulative outcome over heroics, on a team you never see.
Judgment Reps: How PMs Train Decision Quality DeliberatelyJudgment is trainable, but only with reps and a scorecard. The decision log, calibration habit, premortems, and a 4-week plan to train PM decision quality.
Stop Calling Yourself an AI PMThe PM role is fragmenting into AI-native, growth, monetization, platform, and GTM labels. Most of it hedges an org-design decision. Name the outcome you own.
The Product Builder Was Never About CodingLinkedIn replaced its APM program with Product Builder and framed it as generalists. The real shift is collapsing the distance between deciding and making.
Comment With a Frame: The Only Move That CompoundsApplause adds nothing and dunking loses the room. Concede the strongest objection, then add a frame the thread did not have. Two exchanges show how.
The QBR Deck Template That Forces Decisions, Not ApplauseA 9-slide QBR deck template: scoreboard vs plan, decision review, kills, next-quarter bets with confidence and reversibility. Outline plus PowerPoint.
Financial Fluency for PMs: Own the Margin Before It Owns YouFinancial fluency is now a PM job requirement. The five numbers to know cold, the finance working session agenda, and a 4-week plan to own your margin.
The PRD Was Already Dead. Slack Proved It in Public.Slack is running tiny cross-functional squads that prototype continuously instead of committing to fixed roadmaps. The org-chart version of the PRD collapse.
Stop Picking Winners. Design the Tournament.Cheap prototyping did not kill product judgment. It moved the decision from picking the winner to seeding the bracket of what gets prototyped at all.
The Resume Is Dead. Here Is What Replaced It.If you hire PMs off resumes in 2026, you screen for prompting skill, not product skill. What still correlates: published thinking a model cannot compress.
The Exec Update Template: Five Numbers, Three Calls, One AskAn exec update template read in 90 seconds: five-number scoreboard, three decisions with evidence, one ask. Plus the agent prompt that drafts it.
Rank AI Features by Downside Exposure, Not EngagementEngagement rewards features people fiddle with. Rank AI features by downside exposure instead: how fast a silent 20% quality drop would cost you customers.
GTM Is Not Someone Else's Job AnymorePMs are absorbing pricing, activation, and positioning. The PM who stops at requirements is describing half a job in a market that pays for the whole one.
Storytelling for PMs: The Skill That Decides Whose Evidence WinsStorytelling for PMs is not polish, it is how decisions get made. Three frameworks, the retelling test, and a 4-week practice plan to win the room.
Falkster's Messaging Gateway: A Prototype-First Ship StoryInstead of three integrations for Signal, WhatsApp, and Telegram, we bridged through one Matrix homeserver. What building it taught that a PRD could not.
What Competing Against Claude Taught Me About FalksterAn honest gap analysis of Falkster against Claude: memory maturity, workspace surface area, long-running agents, and distribution. The gaps, not a battlecard.
Downside Exposure, Extended: When the Model Is Wrong About YouDownside exposure covered your AI feature failing a user. The second half: when an AI you don't control represents you inaccurately to a buyer.
The First 90 Days as an AI-Native CPOFirst 90 days as a CPO in 2026: audit reality, instrument cost and quality, kill dead ceremonies, and defend the judgment calls only you can make.
The PRD Collapsed Into Three Artifacts (Brief, Prototype, Eval)The PRD collapsed into a one-page brief, a prototype, and an eval. Answering the four objections 21 PMs raised, with copy-paste templates.
The PR/FAQ Template: Amazon's Working Backwards for Your BetsA complete PR/FAQ template: press release skeleton, external and internal FAQ, the narrative meeting protocol, and when not to use it. Markdown and Word.
AI Evaluators Are Your New PM AudienceAI systems now decide if your product gets recommended before a human sees it. Product legibility to a model is a new PM deliverable, not SEO reskinned.
The CEO Who Can't Read Their Own Product AnymoreAgents break the link between dashboards and your product. Numbers stay green while the experience degrades. Three things to read instead of more KPIs.
PM Days 61 to 90: Own One Bet and Kill One CommitmentPM days 61 to 90 in a new job: put your name on one falsifiable bet, kill one inherited commitment, and close the loop with a day-90 note to your manager.
Product Ops Is the Job AI Made NecessaryAs the PM role splits and roadmaps flatten, coordination has to go somewhere. It goes to product ops, quietly the highest-leverage seat, not an admin function.
The Decision Log Template: Train Judgment Like a MuscleA decision log template for PMs: schema, worked examples, the weekly 10-minute ritual, and the quarterly calibration review. Markdown and Excel.
I Ran My Staff Meeting Off a Live Dashboard for a QuarterI replaced weekly staff meeting status updates with a live agent-built dashboard. What it killed, what it surfaced, and the one mistake I would not repeat.
CPO Days 61 to 90: The Kill List, the Bets, and the ReadoutMonth three for a new CPO: build the kill list with the inertia test, kill two things publicly, size 2-3 first bets, and write the SCQA day-90 readout.
The Founder Mode MisreadFounder mode got read as permission to micromanage. The real lesson: stay close to the few decisions only you can make, automate the rest.
PM Days 31 to 60: Ship One Artifact That CompoundsPM days 31 to 60 in a new job: skip the quick win, ship one compounding artifact. Four options, selection criteria, a worked example, and the failure modes.
The Board Deck Product Section: An SCQA Slide KitA 7-slide board product section in 10 minutes: scoreboard, what changed, bets table, kill list, the one risk, and asks. No feature parade. Templates inside.
The AI Product Operating Model Has a Survivorship ProblemThe AI product operating model diagnosis is right. The evidence is every company that lived. Survivorship bias makes a true claim look far more proven.
I'm a Company of One Because I Refused to Vibe CodeI crossed ~$1.5M ARR solo in under two months. It only works because I built agentic engineering, not vibe coding. The substrate does the carrying.
The Cost of Being Wrong Is the Only Number That Matters NowSpeed is free now. The binding constraint for CPOs is reversibility: ship cheap-to-undo decisions instantly, slow down deliberately on the ones you cannot undo.
Kill the 'I Built It in a Weekend' FlexSpeed to demo is a vanity metric. The 'I built it in a weekend' flex measures the cheapest, least durable thing. Ask if it was still running the weekend after.
The Mental Models That Break When Building Gets CheapMost decision frameworks quietly price building as scarce. When AI makes building cheap, four of them invert and start giving the wrong answer.
The Vibe Coding Crossover Point: Google's New SDLC WhitepaperGoogle's whitepaper puts the vibe coding crossover at 3 to 10x. The part that costs you is the quiet graduation, and it is a CapEx versus OpEx call.
The Product Strategy Memo Template (Pyramid Principle)A six-page SCQA strategy memo template after Minto: answer first, three grouped arguments with evidence slots, anti-bets, premortem risks, and the ask.
Five Questions That Make a Product Review Worth Your TimeMost executive product reviews are status theater. Five questions convert them into decision-forcing sessions that surface judgment, cost, and risk.
CPO Days 31 to 60: Instrument the Org Before You Steer ItMonth two for a new CPO: stand up three instruments, build the coalition map, ship one visible decision, and renegotiate the scoreboard with your CEO.
The PM First-90 Kit: Contract, Map, and Day-90 Note TemplatesFour templates for a PM's first 90 days: a one-page manager contract, signal map, decision archaeology worksheet, and the day-90 note. Fill in 10 minutes.
Your First 30 Days as a PM: Build a Signal Map, Skip the Coffee ChatsFirst 30 days as a PM in a new job: run decision archaeology, build a signal map of where truth enters the building, and write the manager contract.
PM Skills 2.0 Is Half a Funeral and Half a BlueprintPaweł Huryn's PM Skills 2.0 review: which of the 68 Claude skills to install (red-team-prd, ship-check) and which half embalms a dying PM role.
Your Next Competitor Has Eight People and Your Whole RoadmapThe most dangerous competitor is eight people with agents: no coordination tax, no legacy, shipping your roadmap faster at a cost base you cannot match.
The CPO First-90 Kit: Ledger, Map, and Readout TemplatesFive templates for a new CPO's first 90 days: listening tour guide, trust and claim ledgers, coalition map, and the day-90 readout. Usable in 10 minutes.
Cagan Built the Church. I Argue We Don't Need One.Marty Cagan's product operating model optimizes for a constraint AI erased. The case for product builders over empowered PMs, and where my thesis breaks.
The Product Budget Is a Compute Budget NowIn an agent-native company, output scales with compute and judgment, not headcount. That breaks how CPOs and CFOs plan.
The CPO Listening Tour Is Broken. Run a Ledger Instead.Replace the 40-coffee-chat listening tour with a listening ledger: nine fixed questions, falsifiable claims, contradiction tracking, and a Friday synthesis.
The Prototype Graveyard: Why Cheap Demos Kill Good IdeasPrototype-first is my thesis, and it has a failure mode no one names. When building is free, the demo replaces the argument and good ideas die in review.
Any Competitor Can Clone Your Feature in a Week. Now What?Any competitor can clone your feature in a week. The moat has moved to distribution, proprietary data, switching cost, and trust. Here's how to audit yours.
Taste Is the Last Moat: Why CEOs Can't Delegate Product JudgmentWhen execution costs nothing, the scarce input is judgment about what to build. Taste is a balance-sheet asset, and the one thing a CEO cannot delegate.
Being AI-First Is a Product Decision, Not a Tooling OneA Twilio CMO drew the line between using AI and being AI-first. Swap marketing for product and the argument gets louder. Here is the product translation.
Product Judgment vs Frameworks: What No Framework TeachesProduct judgment vs frameworks: 20 years at Microsoft, Adobe, and Salesforce taught me what no PM framework can. Judgment comes from being wrong at scale.
AI Isn't Replacing Developers. It's Eroding Them.Michael Lawrence's piece on AI eroding developer craft lands. The 2020 dev would have caught the null check in 12 seconds. The fix is evals, not nostalgia.
IDSD Is Spec-Driven Development With a New Acronym. Kill Both.Intent-Driven Software Development (IDSD) is pitched as the fix for SDD. Same methodology layer, different sticker. The prototype is the spec.
The PM/PO Split Is Broken. Merging Them Misses the Point.Paweł Huryn is right that splitting PM and Product Owner is broken. The fix isn't merging them. It's killing the Scrum frame the roles exist inside.
The 5 PM Websites in Your Bookmarks Need to Become Agents.ProductHunt, Growth.Design, Medium, IndieHackers, TechCrunch. Ankush Panday's daily PM list. In 2026 you don't visit them. Five agents watch them for you.
Yes, Scrum Is Obsolete. The Replacement Already Has a Name.Alex Polyakov's case that Scrum is obsolete lands. He stops short of naming what replaces it. The builder pod: outcome ledger, eval, no ceremony tax.
Three Designer Archetypes in 2026. The PM Split Is the Same Shape.Jenny Wen at Claude hires three designer archetypes; the traditional senior isn't one. The same split hits PMs: builder, specialist, prototype-coder.
The Five-Row Eval Template That Replaced My PRDFive-row eval template that replaces the PRD: Behavior, Input, Expected, Scorer, Threshold. Six failure modes that kill bad evals. Worked example included.
Kill the AI Office Hours. They're 2026's Agile Transformation.AI office hours are 2026's agile transformation: ceremony, no shipped output. Three replacements: paired shipping sessions, eval reviews, kill list reviews.
The PM Interview Question I'd Refuse to Answer in 2026PM interview red flags 2026: three questions that signal a translator PM role, three that signal a builder PM role, and the polite exit script for a mismatch.
Ship Story: The Discovery Week I Ran With Three Agents and No CallsAI agent discovery week, no live customer calls. Three agents (triage, theme synthesizer, opportunity ranker), 20 hours total. Stack, inputs, where it broke.
The 20-Minute Customer Call Triage AgentCustomer call triage agent: 60-minute call to tagged, theme-clustered artifact with a one-line action in 20 minutes. Stack, prompt, eval rubric, batch run.
PMs Should Vibe Code Internal Tools, Not Customer FeaturesVibe coding with Claude Code is the best new skill for PMs. But shipping vibe-coded features to customers is malpractice. Build internal tools instead.
Product Management Isn't Dead. It Escaped the Product Team.Aaron Levie called the emerging role 'agent operator.' The cleaner name: Product Builder plus Agent Operator. Same skill, multiplied across every P&L function.
Six pushbacks I keep hearing on vibe codingSix confusions I hear most when I say vibe coding isn't a strategy: 'demos sell deals,' 'speed beats clarity,' 'evals slow us down.' Here's how I respond.
Customer Discovery When Your Customer Is an AgentContinuous discovery assumes a human customer. When your customer is an agent, six methods apply: agent telemetry, failure-mode interviews, capability audits.
Software Categories AI Is Killing: A PM's Field GuideThree forces are collapsing entire software categories. Which are dead, which are reshaping, which have moats, and what to do this quarter if you build in one.
Distribution Beats Product, and AI Made It HarsherAdobe taught me distribution beats product. I unlearned it at startups for a decade. AI made product cheap to build and distribution the only moat.
Per-Outcome Pricing: What Gets Clearer and What Gets TerrifyingThree SaaS-to-per-outcome pricing redesigns: support automation, sales enablement, contract review. Margin math, dispute mechanics, and contract terms.
The Builder Trap. Field Notes from Two AI Calls This Week.Two calls this week. A CMO buying more sophisticated silos. An engineer building more sophisticated silos. Same trap. What outcome thinking looks like.
From SaaS to Service-as-Software: The CPO's Field GuideThe CPO field guide to Service-as-Software: outcome-based pricing, what changes inside the product, the new org chart, and when to make the shift.
Agile Was a Coping Mechanism. What Replaces It Isn't Waterfall.AI killed Agile's two founding constraints. Right diagnosis, wrong destination: waterfall is not back. What replaces Agile in 2026 is the spec-first loop.
Kill the OKR. The Quarterly Cycle Is Now a Liability.The 12-week OKR cycle is now a liability. Replace it with a Rolling Outcome Ledger: 3-7 active bets, 2-4 week decision deadlines, no end-of-quarter scramble.
What a PM Actually Does All Week, Now and in 2028An hour-by-hour breakdown of the modern product manager's week, and a side-by-side comparison of every task before and after the agent fleet rewires the job.
AI Deletes the Director of Product, Not the PMEveryone asks if AI replaces the PM. Wrong layer. The role automated first is middle management, the Group PM and Director who mostly move information.
Survivorship Bias in AI: Interview the Planes That Didn't Come BackWald's WWII bomber lesson still runs product management. AI worsens the bias, but agents finally make missing data affordable to study. Here's how.
The CFO Conversation: Defending a 20-Point Margin DropThe single most important conversation of the entire pricing migration. The script, the part that almost didn't land, and the seven leading indicators.
Field Report: What Broke When We Killed Our Per-Seat TierMigration drift, an unexpected churning cohort, a unit definition that broke at scale, dispute volume at 12x projection. After-action from the sunset day.
Kill the Feature Request Queue: Your Backlog Is a SignalYour 847-item backlog rationed scarce engineering time. Three replacements: signal synthesis, prototype-first investigation, and an outcome bet ledger.
The PM-to-CPO Bridge in 2026Most PM-to-CPO advice is generic. The 2026 CPO seat demands business-model literacy, agent fleet operations, and public strategic posture. The 12-month track.
Direction Dashboard AgentCompiles seven leading indicators every morning, predicts outcomes 4-8 weeks ahead, runs a weekly Goodhart audit. Direction at AI-native velocity.
Outcome Accountability Is a Luxury Good. Measure Direction.Outcome-driven roadmaps assume 6-12 month measurement cycles. Agents iterate ten times a week. The dual-cadence direction-metric system that closes the gap.
Field Report: 90 Days Inside a SaaS-to-Agents TransitionWhat we believed, what we did, what broke, what we changed. An anonymized after-action report from the first 90 days of a $30M ARR cannibalization.
Continuous Discovery Doesn't Scale for AI-Native ProductsTeresa Torres' continuous discovery is the right answer for human-centric SaaS and the wrong answer for agent products. The punctuated discovery alternative.
Renewal Risk Agent for Migration CohortsCatches the accounts whose operational dashboard says fine but whose qualitative signals say otherwise. Three-dimension scoring, 90 days before renewal.
The Jevons Cliff in Outcome PricingInference costs drop 50% per year. Your $0.50/resolution price is a moving target. Hold or pass? The framework nobody has published until now.
Kill the APM Program. Hire, Train, Promote Builders.LinkedIn CPO Tomer Cohen killed the APM program. The 12-week Product Builder apprenticeship replaces it: four prototypes, one production ship, no rotation.
Margin Watch AgentTracks gross margin per outcome daily, classifies compression cause, and forecasts the Jevons cliff. Pricing reviews stop being post-mortems.
The Pricing Migration Sequence: An 18-Month Quarterly PlaybookPer-seat is dying, but it pays your payroll. The quarter-by-quarter playbook for moving $20M ARR to hybrid outcome pricing without crashing NRR.
Against Product Intuition: Why I Stopped Trusting MineWhy I stopped trusting my product intuition after 20 years. It is pattern-matching on a world that no longer exists, and AI rotates the patterns faster.
Board Narrative Drafter AgentThe most expensive document in product management compiled while you sleep. Quarterly board updates from data sources, comp set, and last quarter's narrative.
The AI Noise Tax. Three Patterns Killing Product Credibility.Three AI patterns I'm done with: LinkedIn 'Claude superpowers' posts, SaaS apps with bolted-on AI sidebars, and PMs claiming prototyping kills empathy.
The Triad Is Dead. Pods Are Dead. Agent Departments Now.Forrester still talks about human-with-agent teams. Wrong unit. Agents form their own departments and interface with human departments on exception.
The Dual Transformation Operating ModelTwo clocks, one product org. Run a legacy SaaS on a six-week rhythm and an agent-native successor on a one-week rhythm without tearing the team apart.
Pricing Migration Tracker AgentWatches every migration account daily, classifies drift type, and recommends a specific intervention. Catch quiet drifters in week one, not week six.
Build a Discovery Agent Stack: Continuous Customer ListeningDiscovery agent stack: seven open-source Claude repos turn customer interviews, support tickets, and public reviews into a continuous loop. Install + recipes.
Build a Measurement Agent Stack: End the Dashboard Hamster WheelMeasurement agent stack: seven open-source Claude repos turn dashboards into a daily outcome digest, a variance detector, and a stakeholder update writer.
Build a Prototype Agent Stack: PRD to Working Demo in a DayBuild a prototype agent stack: eight open-source Claude repos take a PM from idea to working prototype in a day, with TDD, design, and security review.
Meet the Agent Operator Team at Falkster.aiSix new hires at falkster.ai. They manage the AI agents that build the product. They are paid strictly in treats. The team has not missed a single anomaly.
The PM Agent Stack: A Bridge to the Enterprise AI BrainFive-part series for PMs whose company is 6-18 months from an enterprise-wide AI brain. Build the personal PM agent stack today with open-source Claude repos.
10 AI Agents I Built That Failed. The Honest Retrospective.Ten AI agents that failed in production: auto-approve expenses, sentiment classifiers, autonomous pricing, stale-docs RAG. The lessons that stuck after each.
The CPO's Coalition Map: Why You Can't Run a Migration AloneFive seats, ninety days, one playbook. The argument that wins each seat in the executive coalition behind any AI-era pricing migration.
Why "Soft Pivots" Always Become Two Companies in One OfficeEvery CPO running a 'gradual transition' to AI ends up running two companies inside one org chart. The five patterns and how to escape them.
The Cannibalization Playbook: How CPOs Win the AI TransitionThe honest CPO move when AI replaces your SaaS: set a sunset, kill the comp asymmetry, pre-sell the gross-margin trough. The seven-decision sequence.
The Eval-First Product Org: Rebuild Around Quality, Not VelocityRebuild a product org around evals, not velocity. The Quality Spine, Product Builder pods, Economics Unit, Discovery Network. The 90-day rebuild plan.
Kill the PRD. The Prototype Is the Spec.The PRD was a 2015 solution to a 2015 problem. Three artifacts replace it: a working prototype (built in Claude Code), an eval rubric, and a one-page README.
What Replaces the Product Org in 2028: Ten PredictionsTen checkable predictions for product orgs by July 2028: median team size six, evals the highest-status skill, hybrid pricing mainstream, APM programs extinct.
200 PM Job Descriptions Reviewed: 98% Hire for a Dead Job200 PM job descriptions audited. 196 could have been written in 2022. The four tells of a dead JD, four tells of a live one, and a before-and-after rewrite.
The PM-to-Engineer Ratio Is the First Thing AI BreaksThe PM layoffs of 2025 and 2026 are not a skills story. They are a ratio correction. When each engineer ships 3x more, the math that sets PM headcount changes.
39 PM AI Agents Deployed: What Stuck, What Died, and WhyAn honest accounting of 39 PM AI agents across 4 product orgs in 80 days. Stage skew, cadence patterns, and the failure mode I kept repeating.
The Agent Sandbox (PM Version): A Complete User ManualThe agent management system I built to run falkster.ai, now public. Nine views, 39 agents, three workflows, a 4-layer brain. See what an AI-native fleet does.
The CPO Mandate 2026: What Boards Expect From ProductThe 2026 CPO board deck: outcome ledger, per-outcome unit economics, eval scorecard, agent inventory, cycle time, headcount-to-output, and willing-to-be-wrong.
The Seven-Agent Reset: A Lean PM AI Fleet, One Agent Per StageStarting over: the seven-agent PM AI fleet I'd build today. One per PM OS stage, each with named signals, a single surface, and a quarterly kill-switch test.
My Best Product Decision Was a Kill, Not a LaunchKilling a product feature is the rarest leadership skill. The best decision of my career was a kill, not a launch. The industry rewards exactly backwards.
Empowered Product Teams Were a ZIRP FeatureThe empowered product team was a ZIRP luxury, not a timeless truth. With capital expensive and AI shifting the constraint, the orthodoxy is breaking.
Auto Bugfix Agent: Zendesk Ticket to Reviewable PR in 11 MinutesAn AI agent that reproduces customer-reported bugs, locates the broken code, writes the fix, adds a regression test, and opens a reviewable PR.
Instant Prototype Agent: Customer Request to Prototype in MinutesAn AI agent that turns customer feature requests into working prototypes. Deploys a preview URL, opens a PR, files a Linear ticket, drafts a Notion doc.
Launch Comms Agent: Six Channels Generated on Every ReleaseAI agent that generates website copy, LinkedIn post, customer email, in-app banner, changelog, and X thread when a feature ships. Six channels, four minutes.
Signal-to-Ship Cycle Time Agent: Measure PM Velocity Across 7 StagesAn AI agent that tracks PM cycle time across the 7 stages of the Product Operating System (Sense to Amplify) and surfaces the weekly bottleneck.
pragmatic-pm-ai-skills Review: A Claude Skill That Tightens PM OutputReview of Valentin Brain's pragmatic-pm-ai-skills: a Claude Skill library that makes PM-AI output incentive-aware, evidence-tied, and free of generic advice.
The Ten-Day Dev Loop: Three AI Agents Collapsed Our 8-Week CycleHow three AI agents (Instant Prototype, Auto Bugfix, Launch Comms) turned a customer Slack DM into a shipped, announced feature in 10 days instead of 8 weeks.
KPI Watchdog Agent: Catch Metric Drops and Ship a Fix PrototypeA watchdog agent monitors your KPIs hourly, investigates the root cause of any drop, and ships a working prototype fix before you finish your morning coffee.
The New Org Chart for AIAI coding tools boosted developer output but org velocity stayed flat. Three fixes: engineer adoption, faster review process, and flattening coordination.
How PMs Ship Their First Pull Request With AI ToolsFour-level PM PR ladder, PLANNING.md templates, PM PR review skill file, and CTO pitch docs. Built from real coaching engagements across B2B SaaS orgs.
Your Product KPIs Are Your Job, DailyA product builder without a daily KPI ritual drives with the rear-view mirror. Pick three KPIs, watch them every morning, elevate to OKRs when they slip.
Mob Prototyping: One Day, One Room, One PrototypeOne day a week, your PM, designer, and engineer build a working prototype together. Cheaper than spec-driven handoffs. Better output than solo prototyping.
Product Builder Job Ladder: Four Fork-Ready JDs from L4 to PrincipalFork-ready job descriptions for Builder PMs at four levels: L4 to Principal. Scope, architecture, mentorship, and experience floors calibrated per level.
The PM Role Is Being Rewritten. Are You Rewriting Yourself?AI-forward companies are hiring for 'Product Builders.' This isn't a quirky experiment. It's the new standard. Here are the six skills that actually matter now.
Build a PM Second Brain from Meeting RecordingsPMs forget 90% of meeting context within a week. How to record, transcribe, embed, and query meetings into a searchable knowledge base that compounds.
I Gave My AI Agents a Performance Review. Three Got Fired.If an agent does the work of a team member, manage it like one. The scorecard, the coaching loop, and the firing criteria for the agents on your product team.
Stakeholder Communication AgentGenerate tailored updates for different audiences: exec summaries, board updates, investor reports, team briefings. All from the same data.
Retrospective Synthesis and Learning AgentExtract learnings from sprint retros automatically. Update playbooks, surface patterns, and drive continuous improvement.
Win/Loss Analysis AgentBi-weekly analysis of won and lost deals. Extract product insights from sales outcomes. What's winning business, and what's holding you back?
Feature Adoption Tracking AgentDaily adoption curves for new features. Identify stuck cohorts and recommend interventions before adoption stalls.
Automated PRD GeneratorConvert prioritized opportunities into PRDs automatically. Drafts based on research context, design specs, and technical requirements.
Automated Release Documentation AgentAuto-generate release notes, customer comms, and doc updates as features ship. Saves hours on documentation and keeps comms consistent.
Tech Debt Impact and Prioritization AgentWeekly analysis of tech debt impact on velocity. Which debt items actually matter? What's the priority order to reclaim speed?
Testable Assumptions Tracker AgentConvert opportunities into testable assumptions. Track validation status weekly. Know which assumptions are holding up your roadmap.
OKR Progress and Prediction AgentDaily OKR tracking with outcome scoring. Weekly confidence predictions for OKR achievement. See what's on track and what needs intervention.
Opportunity Prioritization and Synthesis AgentWeekly synthesis of all your DISCOVER agent outputs into a prioritized opportunity stack. OST-ready opportunities with impact estimates and dependencies.
Automated Sprint Planning AgentConvert prioritized opportunities into user stories and sprint plans every Monday. Accounts for team capacity, tech debt, and dependencies.
Customer Segmentation AgentWeekly updates to your customer segments and personas. Cohort analysis, segment evolution, and behavioral groupings - automatically maintained.
Customer Interview Synthesis AgentWeekly automated synthesis of your customer interviews. Key themes, hypotheses, and actionable insights - without the tedious manual work.
Automated Customer Journey MappingBuild rich customer journey maps bi-weekly from session replays, support patterns, and research data. See exactly where users get stuck.
Every Type of PM Needs a Different Agent StackTechnical PMs, Growth PMs, Platform PMs, and Product Ops have different jobs. Their AI agent stacks should be different too. Here's what works for each role.
Every CPO Job Is a Turnaround: The First 90 DaysNobody hires a CPO when things are going well. Every CPO role is secretly a turnaround. What the first 90 days actually require, in the AI-native era.
NPS and CSAT Deep Dive AgentDaily and weekly analysis of your NPS and CSAT scores. Segment breakdowns, feature drivers, and what's actually making your customers happy or frustrated.
Automate Support Pattern DetectionA daily agent that extracts signals from your support queue: which problems are rising, which segments are hurting, and what real trend lies beneath the noise.
Claude Skills Every PM Should Build TodaySkills are the most underused feature in Claude. Here's how to build five that genuinely improve your daily PM workflow - with exact prompts and templates.
Founders Don't Want a CPO. Here's What They Want.The CPO founder relationship fails for one unspoken reason: founders want their product judgment cloned and scaled without losing control. What I learned.
Build Your Market Intelligence AgentBi-weekly deep dive into your competitive landscape. Win/loss analysis, market shifts, feature parity, and strategic recommendations so you're never surprised.
Agent-to-Agent Dispatch: The Product Org Chart Nobody Is DesigningEveryone is building single AI agents for PMs. The real shift is agents handing work to other agents, with the PM as dispatcher. Here is the architecture.
The PM Automation Stack: OpenClaw vs Claude vs Everything ElseOpenClaw went viral. Claude Code went deep. n8n went wide. What each tool actually automates for PMs, what's vaporware, and the exact setup that works today.
How a 2-Hour Prototype Killed a 3-Month ProjectFive customer calls and a two-hour prototype showed our feature design was structurally wrong. How prototype validation saved 600 engineering hours.
Build Your Release Checker AgentThe final gate before you ship. This Thursday agent verifies QA, docs, GTM materials, and sign-offs, giving a clear go/no-go for each feature.
Two Weeks of Agent Tuning: What I LearnedMy first agent reports were 60% noise. After two weeks of calibration, they became the first thing I check every morning. Every adjustment I made and why.
From Zero to Weekly Discovery in 30 DaysWe went from 'we talk to customers quarterly' to 'every PM runs weekly discovery calls.' It wasn't a mandate - it was a system. Here's the 30-day playbook.
The PM-as-Translator Is DeadFor 15 years, PMs were professional middleware reformatting customer input into decks for engineers. AI just stripped away every translation layer.
What Happened When We Killed Our Most-Requested Feature200+ customer requests. Sales demanding it. The CEO mentioned it in an all-hands. We killed it anyway - and activation improved 15%. Here's why.
Build Your Daily Red Flag AgentThe first agent every PM should deploy: a morning scan of support, backlog, Slack, and production that tells you exactly what needs your attention today.
How I Caught a Churn Signal 3 Weeks EarlyA Red Flag agent caught a 400% support spike on a tier-1 account. Three weeks later they renewed at full terms. Here's exactly what happened.
Product Management Without Revenue Made Me BetterThe highest-stakes product of my career had no revenue model. Mission-driven product management forced pure outcome thinking, the way AI agents force it now.
The PRD Is Dead. Here's What Replaces It.Product Requirements Documents were a necessary evil in a world where building was expensive. That world is gone.
The Quarterly Business Review Template That Gets You PromotedA QBR template for product managers: 5-section structure, the metrics leadership cares about, narrative framework, and how to land a clear ask. Free download.
Build Your Product Health AgentDaily 4 PM pulse check: engagement, activation, performance, and support trends synthesized into one story. Know exactly where your product stands each day.
Setup Guide: Connecting Claude to Your Data SourcesStep-by-step guide to wiring Claude into your actual work tools via MCP. Based on my production setup at Smartcat.
Setup Guide: Self-Hosted Agents with OpenClawFor teams that need on-premise data control. Same agents, same data sources, self-hosted infrastructure.
The AARRR Dashboard You Actually NeedForget vanity metrics dashboards. Here's how to build a pirate metrics dashboard that tells you where your product is leaking value - and what to fix first.
Build Your Competitive Intelligence AgentStop manual competitive research. This agent monitors competitors, tracks customer mentions, and delivers a weekly intel report so you always know your market.
20 AI Prompts That Replace 20 Hours of PM WorkCopy-paste AI prompts for product managers into Claude or ChatGPT. Tested across discovery, roadmapping, retros, and stakeholder updates. Save 20 hours a week.
The Guerrilla PM PlaybookNo research team. No data analyst. No design support. Here's how to run world-class product management anyway with AI, scrappiness, and 2 hours a week.
OKR Writing That Doesn't SuckMost OKRs are disguised task lists. How to write outcome-focused OKRs that drive product decisions, with templates, anti-patterns, and real examples.
Build Your Team Triage AgentYour #team-product channel is a firehose. This agent reads every message, categorizes issues, assigns owners, and surfaces what's unanswered, twice daily.
Empowered Teams When Your CEO Doesn't Buy InYour CEO wants feature roadmaps. You want empowered teams. Here's the guerrilla playbook for building autonomy from the inside without getting fired.
Prompt Logs Are the New Switch InterviewStop scheduling discovery calls to learn what customers want. They are already telling your AI in plain language. Mine the prompt logs instead.
Build Your Engineering Capacity AgentEngineering capacity agent: flags overloaded engineers, catches PTO gaps weeks early, and tells you if the roadmap won't fit before the sprint starts.
Running Your First Product TrioProduct, Design, and Engineering making decisions together weekly, not in a meeting, but as an operating rhythm. How to start even if your org doesn't do trios.
PMs Won't Be Automated - They'll Be AmplifiedThe hot take that AI will replace PMs is wrong. But the PMs who survive will look nothing like the PMs of today.
Build Your Documentation Gap AgentFeatures ship without docs and customers find the gaps. This agent scans tickets and the release calendar to catch documentation gaps before they become fires.
Shipping Velocity vs Being Right: My Two Years Shipping ZeroMicrosoft Research paid me to ship nothing for two years. It was the best product training I ever got. Shipping velocity is not the same as being right.
The AI-Powered Weekly Review: A System for PMs Who ShipA step-by-step weekly review system that uses AI agents to gather data, surface insights, and prepare your week - in 30 minutes instead of 3 hours.
Build Your Weekly Ops Digest AgentDaily agents catch fires. This weekly digest spots the trends: recurring issues, degrading resolution times, and systemic problems that daily reports miss.
Build Your Product Ops AgentDaily product ops report ranked by ARR at risk: top complaints, feature gaps, usage anomalies, and at-risk customers synthesized from five data streams.
Build Your Daily Focus AgentYour AI chief of staff: reads your calendar, scans Slack, checks the roadmap, and delivers the 3 things that actually matter today before your first meeting.
How to Build a Working Prototype in 60 Minutes (Step-by-Step)A concrete walkthrough of building a real prototype with AI - from problem statement to deployed, testable app. With the exact prompts and workflow.
Build Your Customer Commitment AgentSales promised a feature by Q2. CS said 'next month.' This agent tracks every commitment, cross-references the roadmap, and flags what is overdue.
Why Acquisitions Fail: The Integration AutopsyWhy acquisitions fail at integration: I watched it three times at Salesforce. The acquired team loses its constraint, inherits parent process, stops shipping.
Put Stakeholder Updates on Autopilot: The Complete Setup GuideHow to build an AI agent that generates weekly stakeholder updates, project status reports, and exec summaries - so you never write one manually again.
Build Your PM Issues AgentCatch slipping features, overdue customer promises, and cross-team dependency failures before they become fires. Scans roadmap and commitments every morning.
Build Your Product Health Dashboard AgentGo deeper than daily metrics. Weekly agent analyzing feature adoption curves, cohort retention, engagement depth, and performance trends to shape strategy.
Build a Customer Feedback Pipeline in One AfternoonConnect Zendesk, Slack, and app reviews into a single AI-powered insight engine. Step-by-step setup guide with code snippets.
Jobs to Be Done Has a Half-Life Now (And AI Is Shrinking It)JTBD assumes the underlying job is stable. AI breaks that assumption. Why the job your customer hired your product for is decaying faster than your roadmap.
Build Your Release Readiness AgentPrevents launch disasters. Checks PRDs, release notes, feature flags, and GTM readiness every Wednesday so nothing ships without proper preparation.
The One-Pager That Replaced Our PRDs (Template Included)The exact one-page template we use instead of PRDs. Covers problem, evidence, solution, metrics, and risks - nothing else.
Build Your GTM Release Monitoring AgentTrack if features are ready to sell, support, and succeed. Monitors discovery compliance, beta standards, and GTM materials daily across five readiness gates.
Leading a Product Team Through Layoffs: What SurvivesLeading a product team through layoffs: the cut is not the hard part. Keeping the survivors building is. What I learned running product through three.
What Happens After Acquisition: My Product Was KilledAfter acquisition your product often dies, and that can be correct. Microsoft killed the product I built. Here is why it was the right call.
Set Up a Competitive Intelligence Agent in 30 MinutesBuild an AI agent that monitors your competitors across 8 sources and delivers a weekly brief to Slack. 30-minute setup, no code, runs forever.
Build Your Roadmap Progress AgentYour roadmap says in progress but engineering hasn't touched it in weeks. This agent cross-references your backlog against actual dev activity daily.
Build Your Executive Report AgentAuto-generated Monday leadership brief: roadmap status, top wins, key risks, and metrics in a 3-minute read. Your CEO walks in already prepared.
Run a 15-Minute Sprint Retro That Actually Improves ThingsAn AI-assisted retro format that gathers data before the meeting, identifies real patterns, and generates action items that stick.
What Acquirers Actually Buy in a Startup AcquisitionWhat acquirers actually buy is not your product or revenue. I sold MVC to Microsoft. They buy a capability, a team, a defensive block, or time.
The Reorg Is a Product Decision Nobody Calls OneOrg design product strategy and Conway's Law: your reorg ships more than your roadmap. Why leaders treat structure as HR when it is product.
Startup Exit Lessons: My $6.5B Win Taught Me NothingMy biggest startup exit was mostly timing and luck. The failure that returned almost nothing is where the real product-leadership learning lived.

2024

Your 'AI Agent' Is Probably Just a Cron JobMost things marketed as AI agents are workflows or automations with a chatbot bolted on. Here's how to tell the difference and why it matters for how you build.
Many AI Agents Are Actually Workflows or Automations in DisguiseAutomations, AI workflows, and real AI agents behave differently. Knowing which you have prevents inflated expectations and missed opportunities.
The AI Revolution Is Faster and Deeper Than the Industrial RevolutionThe AI revolution compresses centuries of change into decades. Here's how companies and SaaS providers can adapt before it's too late.
Is Mob Programming the Right Approach for Your Team?Mob programming promises collaboration and quality, but at what cost? A look at why fully empowered Scrum teams are almost always the better bet.
The AI Product Engineer: One Person Doing What Used to Take a TeamAI is blurring the lines between PM, design, and engineering. People who work across all three with AI tools will own the next decade of product.
The Rise of the AI Product EngineerAI is enabling one person to own the full product lifecycle. The AI Product Engineer blends PM, engineer, and designer skills - all amplified by AI.
The Evolution of Product Management Over the Last 20 YearsFrom project coordinators to strategic leaders to AI-powered product engineers. A 20-year journey through PM's transformation and what comes next.
The Shift from SaaS to Service-as-SoftwareSoftware is evolving from a tool humans use into a workforce that does the work. Here's how the SaaS to Service-as-Software shift changes everything for PMs.
SaaS Is Becoming Service-as-SoftwareSoftware used to help people do work. Now it does the work itself. This shift changes how you build, price, and think about product. I'm living it at Smartcat.
AI Agents and the Future of Work: A Pixar-Inspired JourneyA narrative-style exploration of what happens when AI agents become your coworkers - told through a Pixar-inspired story about Casey and the future office.
When Your Data Becomes the ProductMany nonprofits offer free services but monetize the insights from your data. With AI making deeper analysis possible, the stakes are higher than ever.
How to Avoid Survivorship Bias in Product ManagementThe British bomber study teaches us that the most critical data is often what we can't see. Here's how survivorship bias shows up in PM and how to fix it.
You're Only Listening to the SurvivorsSurvivorship bias is quietly wrecking your product decisions. You're building for the users who stayed and ignoring the ones who left. Here's how to fix that.
Why Users Don't Know What They Want Until You Show ThemFeature requests are reactive. Visionary product development means anticipating needs users can't articulate yet - and showing them what's possible.
Show, Don't Tell: Why Users Can't Describe What They NeedFeature requests tell you what's broken. Prototypes tell you what's possible. I stopped treating user feedback as a roadmap and started using it as input.