Articles on scaling,
AI and risk.

Frameworks from real engagements on scaling, organizational risk, AI and decision-making, written from inside the work.

Jun 17, 2026 19 min read

The assessment is the product

Why a clear diagnosis beats an open-ended retainer. How an organizational assessment prevents the expensive blind mistakes - unnecessary layoffs companies later reverse, key-person and knowledge loss, decision bottlenecks, and brittle architecture - grounded in Gallup, SHRM, Gartner, and Forrester data. What you get at the end, and how to act on it.

Read the essay

Scaling a company? Start with these three.

1Read in orderWhy problems get more expensive as you growWhy catching problems early costs 10x less.12 min
2Read in orderThe assessment is the productWhat the diagnosis actually delivers.19 min
3Read in orderWhy AI changed the math on senior hiringThe smart way to staff what comes next.17 min

Every article, one index.

29AI Strategy · Operations · SoftwareWhere AI Still Needs Deterministic Rules: At Work, In Software, As a CustomerAt temperature 0, 1,000 identical requests returned 80 different answers in one test, and underwriters priced identical cases 55% apart. AI models are built to vary: useful for a draft, costly for a decision. Where rules and code must still decide, the multiple-choice and other fallacies that hide it, what it means for UI and UX, and a short checklist for the line between what the model proposes and what the rules decide.Oct 613 min
28Organizational Design · Change ManagementWhy People Are Right to Be Skeptical of Organizational ConsultingThe famous claim that 70% of change initiatives fail has no empirical basis - a peer-reviewed review traced it across five sources and found none (Journal of Change Management, 2011). HBR still reports 60-90% of strategic plans never fully launch. A model is not a company. Why generic frameworks don't survive contact with a real business, a real example (a civil-engineering client who, after three hours of listening, needed a note-taker instead of a consultant), and why building the fix yourself - not just recommending it - changes what you can propose.Sep 3011 min
27Organizational Design · ScalingWhy Fast Growth Breaks Companies Without the Right StructureA study of more than 3,200 startups found 70% scaled prematurely and 74% of failures traced back to it, while no prematurely-scaled startup ever passed 100,000 users. Success hides structural damage longer than failure does, because revenue climbing looks like proof the current way of operating still works. Why the fix is not a generic scaling playbook, but structure built to the company's own identity - its culture, market, stage, and resources.Sep 2713 min
26Productivity · Decision ReviewWhy companies keep adding and rarely review past decisionsWhere AI frees up hours, most of them go into another feature, another workflow, another agent. Four researched mechanisms explain why: people think of adding before removing (Nature, 2021), the person who made a failing decision invests the most in it (Staw, 1976), people bid 63% more for an IKEA box they had assembled themselves, and 693 participants reasoned more wisely about someone else's problem than their own. What a team can do on its own, when a review from outside is the better tool, and where that review has limits.Sep 199 min
25For Investors · Due DiligenceOperational due diligence: find the problem while the fix is still cheapAn operational problem starts in how people work, hardens into the technology, and only then reaches the numbers. Financial diligence sees it when the fix is capital, a technical review when the fix is a rebuild, operational diligence while the fix is still a process change. Same problem, three prices. Why it sits alongside financial and technical diligence, why it is the cheapest of them, and which investors it fits best.Aug 1713 min
24Productivity · PrioritiesWhy work that feels productive often moves nothingPeople and companies rarely avoid work. They pick the work that cannot reject them - another article, a bug in a feature nobody opens, the next roadmap item instead of the refactor. Only 6.4 of every 100 launched features drive 80% of usage, and elite goalkeepers dive 93.7% of the time against the odds for the same reason. Includes my own three months of excellent work that sold nothing, what AI did to the signal, the strongest objection to the whole idea - that below-line work is how capability gets built - taken seriously rather than waved off, and the identity-first response: plan for who you are now, not who the plan assumes you are.Aug 1719 min
23KPI Ownership · Role DesignWhy your most important number has no ownerA KPI is defined by an outcome and a role is defined by a title, so cascading a number down a title-based chart lands it on nobody. Three artifacts fix it: a Contribution Charter for what a team owes the company, a Coverage Map showing which capabilities are held by whom, and a Contribution Ledger that measures a person by what they moved, hold, decide and multiply. Includes where a pure need-based structure fails, and the Israeli equal-pay constraint that keeps the title alive.Aug 1715 min
22Hiring · Role DesignWhy the job description does not match the jobCompanies bundle five jobs into one role because a role is the only unit they know how to purchase, then select for it with the weakest method in the table. Personnel-selection research finds unstructured interviews predict performance worse than structured interviews and work samples, though the 2022 revision shows the gap is smaller than the 1998 estimates suggested. What the bundle costs candidates, what it costs companies in headcount they never needed, and what AI changed (volume) and did not change (accuracy). Companion to The GEAR Model.Aug 1014 min
21Leadership · Decision MakingWhat a CEO is actually asked to beWritten honestly, the founder role demands ten traits at once: original thinking, focus, self-generated motivation, leadership, stress tolerance, the willingness to reverse direction publicly, risk appetite, trusting your own read, permanent openness to criticism, and doing it all with nobody above you. Two of them contradict each other, and almost nobody holds all ten. Inside a company the weak ones get absorbed by other people; alone at the top nothing absorbs them. Which traits actually fail, how to find your two, where an agent can hold a specific gap, and why staying employed is the right answer for a lot of people.Aug 714 min
20Interim and fractionalGet the senior work done without making the hireWork arrives on a Tuesday; headcount arrives a budget cycle later. What the on-demand model actually is, how it differs commercially from a recruiter, a freelancer, and an agency, the full cost stack of a permanent hire with the arithmetic showing, and the 8 KPIs to hold it accountable - including the one that should eventually tell you to stop and hire someone permanent.Aug 514 min
19Greiner Growth Model · Organizational ImprintingThe Growth Crises You Can See ComingGrowth doesn't break companies - the predictable transitions do. A field guide combining the Greiner Curve (the six growth walls) with organizational imprinting: each crisis is also a window when culture is soft enough to re-shape, so you can imprint the habits that prevent the next wall - cheaply and self-enforcing. With the delegation use case and its monetary value. Grounded in Greiner, Stinchcombe, and Marquis & Tilcsik.Jul 2813 min
18Decision-Making · MeasurementHow to measure what good planning is actually worthGood decisions are invisible - nothing breaks, so nobody counts them. A measurement model for the ROI of planning well, across three ledgers (Build, People, Alignment), with a scorecard and money formulas you can run on your own numbers. Grounded in Bain, McKinsey, DORA, Gallup, SHRM, Brandon Hall, Pendo, and NIST.Jul 2717 min
17People Operations · Organizational DesignRunning HR after you flatten the orgFlatten the org and the people system becomes homeless unless someone owns it. How the HR function runs as an enabling team across the full lifecycle - hiring, onboarding, leveling, performance, recognition, and letting go - plus the honest failure modes of going flat. The people layer underneath the GEAR model.Jul 2412 min
16Organizational Design · Operating ModelDesign your org around what it needs, not the job titlesMost org charts optimize for control. The strongest ones optimize for what the business needs: to sell more, spend less, move fast, and stay adaptive (Growth, Efficiency, Agility, Resilience). A needs-first operating model with an org chart, a decision-and-delivery flow, and a six-step rollout you can own. Grounded in Bain, McKinsey, Team Topologies, DORA, and Galbraith.Jul 2313 min
15Organizational Design · Hiring · ResearchWhy some roles cannot be filled by one personSome roles are hard to hire for not because talent is scarce, but because they demand two opposite minds - assessment and locomotion - in one person. Product Manager is the textbook case. The research (regulatory mode, the Rubicon model, effectuation, Rocket Fuel), what organizations actually pay for, which roles fit each mode, and three ways to redesign the roles that demand both: split, pair, or scope to the phase.Jul 209 min
14Future of Work · Fractional Leadership · AIWhy AI changed the math on hiring senior peopleAI did not make senior operators obsolete - it made them leveraged enough that buying a full-time seat is now the expensive option. Why the fractional model is moving from stopgap to default, the real cost comparison against hiring, an honest account of when fractional is the wrong choice, and four tests for evaluating an operator. Written by someone who runs her own consulting practice as a one-person, AI-powered company.Jun 1317 min
13Risk Management · AI Operations · FrameworkWhat to move fast on, and what to slow downSpeed is not the default virtue of the AI era - judgment is. Three verified case studies (CrowdStrike, Air Canada, Knight Capital) show what happens when velocity outpaces readiness. Includes the Velocity Risk Matrix: a two-axis framework for classifying every process by reversibility and blast radius, and five categories that must stay permanently slow.Jun 220 min
12Product Management · Future of Work · OpinionWhy product managers should own the company's AI knowledgeThe PM is absorbing four roles - Research Director, Project Manager, Designer, and with the right developer infrastructure, feature development itself. Simultaneously becoming the knowledge architect every AI system depends on. Grounded in DORA metrics, SPACE framework, Teresa Torres, Marty Cagan, and Karpathy's vibe coding. Includes a provocative 2027 PM job description and four counterarguments worth fighting about.May 2920 min
11Practical Framework · MeasurementShip KPIs like featuresA practical method for adding new KPIs with the rigor product teams use for features. Built on the Three Pillars (Outcome / Execution / Foundation). Includes a step-by-step AI-native implementation using Google Workspace + Claude - one operator, 90 minutes, end-to-end (define, collect, calculate, track, feedback, update, communicate) - plus a dedicated security, compliance, and cost-risk chapter with a 12-item pre-launch checklist. Grounded in Parmenter, Grove, Doerr, Cagan, Kaplan & Norton, Sean Ellis. Includes a worked example and a poker case study.May 2426 min
10Critical Essay · Future of WorkWhat AI will not fix, and what it will reveal95% of GenAI pilots fail. 67% of AI adopters work more hours, not fewer. 60% of agentic projects fail this year for one reason: bad data. The structural problems AI can't fix, the new ones it's creating, the stacked migration nobody's talking about, and 7 predictions for 2027-2030.May 2318 min
09Critical Essay · Risk ManagementWhen your measurement stack becomes the riskGoodhart's Law, Wells Fargo's $3B scandal, the UK PPI crisis, why CSAT is expensive theater, and how isolated departmental KPIs create internal civil wars between Risk and Product. With research and a 5-principle framework for healthier KPI architecture.May 2211 min
08Economics of PreventionWhy organizational problems get more expensive as you growEvery problem you don't catch now costs 10x more at the next scale. The math behind prevention, the three categories of findings (leaks, bottlenecks, risks), and when to assess.May 1312 min
07AI TransformationEnterprise AI Transformation: The Operator's Playbook for Regulated Industries5-stage framework to move organizations from AI experiments in engineering to AI as operating model. Built for banking, fintech, insurance, healthcare. Operator-led, vendor-neutral.May 1014 min
05For InvestorsPre-Investment Due Diligence: A Builder's-Eye ViewWhat documents and financials do not show about a target company, and how an operator assesses execution readiness in five phases. Not investment advice.Apr 2511 min
01AI & BusinessAI Readiness Assessment: The Complete Guide for SMBs in 20265-dimension framework, step-by-step process, and a scoring table to see where you stand and what to fix first.Apr 1114 min
06My MethodHow I work with a company, start to finishThe four-phase method I use to diagnose, design, and deliver organizational change - drawn from 10+ years leading regulated tech and hands-on AI work.Feb 289 min
02CareerWhy I Left a 10-Year Tech Career to Become an Independent ConsultantThe honest version: why I left, what I underestimated, and what I learned about consulting in the first 6 months.Feb 149 min
03LeadershipWhat NVC Taught Me About Leading Teams Under PressureNotes from a Marshall Rosenberg workshop. The four-step framework that transforms how leaders give feedback and navigate conflict.Mar 288 min
04PerformanceHow high achievers break through a plateauWhy working harder stops working - and what to do instead. A practical framework for getting the most from yourself.Mar 1410 min

General note

Have a real challenge to work on? Frameworks are useful. Working through a specific problem with someone who's been there is more useful. If something here resonates, let's talk about your situation directly. Send a brief

Articles on scaling, AI and risk.

The assessment is the product

Scaling a company? Start with these three.

Every article, one index.

Articles on scaling,
AI and risk.