evolving . work

IT outsourcing · coding agents · service management 9 min

When Coding Gets Cheaper, the IT Handoff Gets Harder to Justify

If your supplier uses coding agents while you still specify and verify the work, it is time to reconsider what you are outsourcing.

№078 · Oct 1, 2026

coding agents · software development · abstraction 9 min

From Writing Code to Managing Agents

Coding agents let developers work with larger units than individual lines. The skill is knowing how to delegate, check, and investigate.

№077 · Sep 29, 2026

AI agents · productivity 11 min

Ten Agents, 250x the Tokens

Forecasts of 24x token-consumption growth by 2030 are not aggressive, they are too low, because they model usage as if it grows linearly per user.

№076 · Sep 26, 2026

AI agents · productivity 13 min

The Myth of the 50-Agent Founder

The fantasy of a solo founder running a company with 50 agents misunderstands what agents do to cognitive load.

№075 · Sep 24, 2026

When Time Collapses, What Breaks First?

Time was an invisible governance system in organizations, not just a constraint.

№074 · Sep 21, 2026

AI Does Not Eliminate Bottlenecks. It Moves Them.

AI does not eliminate bottlenecks.

№073 · Sep 19, 2026

local AI · Qwen · benchmarks 9 min

Splash! A Test of Real Work

Two Qwen 3.8 27B builds audited a wine e-commerce site. Splash finished sooner, but their paths diverged. The session logs show what we can compare.

№072 · Sep 19, 2026

local AI · AI models · Apple Silicon 45 min

The Best Local Models for a 64 GB MacBook Pro

The page I keep updated with the local model I would actually run on a 64 GB MacBook Pro today, why I picked it, and what else I tried.

№071 · Sep 19, 2026

Models

Jev and the Renaissance of Classifiers

Jev, its open-source alternatives, and the emerging decision primitives that could automate parts of software development.

AI models · classifiers · automation 11 min

№070 · Sep 18, 2026

local AI · Qwen · benchmarks 23 min

Battle of the Qwens: 27B Versus Custom Flash

I built a custom Qwen Flash model to beat the 27B I rely on. It is quicker on the work both of them finish, and it writes well — but sometimes it never stops.

№069 · Sep 18, 2026

The More Abstract the System, the Harder It Is to Fix

Abstraction is how organizations scale: metrics replace direct observation, processes replace individual judgment, dashboards replace frontline presence.

№068 · Sep 17, 2026

local AI · Qwen · quantization 27 min

Which Qwen 3.8 27B Build Should You Run?

Half a dozen downloads all call themselves Qwen 3.8 27B. I compared them on a 64 GB MacBook Pro to find the one worth keeping, and when the others are worth it.

№067 · Sep 16, 2026

local AI · Apple Silicon · AI models 25 min

Running Local AI Models on a 64 GB MacBook Pro

What a 64 GB MacBook Pro can actually do offline, and how I choose a local AI model package, a runtime and the memory settings, tested mainly on coding and structured tasks.

№066 · Sep 15, 2026

local AI · Qwen · quantization 25 min

How I Made Qwen Flash-Next Fit on a 64 GB MacBook Pro

Qwen's Flash-Next ships bigger than my Mac's memory. I cut the expert bank, mixed the precision and left the biggest file on the SSD. The build runs, with limits on memory headroom and task completion.

№065 · Sep 15, 2026

The Correctness Economy

AI made volume free; correctness became the binding constraint.

№064 · Sep 13, 2026

The Smooth Frontier

The jagged frontier that defined AI capability in 2023 was not an intelligence problem.

№063 · Sep 10, 2026

AI agents · productivity 12 min

The Dangerous Illusion of Giving Agents Access

Giving an AI agent access to email, CRM, documents, or payments is not a productivity decision.

№062 · Sep 7, 2026

The 17% That Matters

AI reasoning and output operate as semi-independent processes: a model can identify the correct answer in its reasoning and then produce a recommendation that contradicts it.

№061 · Sep 4, 2026

Why These Models Fail Like Humans and Not Like Software

AI failures are not nonsense — they are near-sense, persuasive while subtly wrong.

№060 · Sep 2, 2026

AI Does Not Think Like Software

Language models are probabilistic systems being deployed where deterministic logic belongs.

№059 · Aug 30, 2026

Models

A Reasoning Model Is a Simulator of Thought

Modern reasoning models simulate what reasoning looks like well enough to produce real reasoning behavior — but without the guarantees of formal reasoning.

№058 · Aug 27, 2026

Models

Why the Best Model Depends on the Shape of the Work

Model selection still happens as if intelligence were one-dimensional: find the strongest score on a leaderboard, deploy it everywhere, optimize later.

№057 · Aug 24, 2026

Centralized Intelligence vs Orchestrated Intelligence: Why Your Model Choice Is Really a Systems Choice

Frontier models are not just different in quality — they express fundamentally different bets on how intelligence should be organized.

№056 · Aug 22, 2026

Want Some Python With Your Burrito?

Companies are stuffing general-purpose intelligence into single-purpose systems and calling it innovation.

№055 · Aug 19, 2026

Is YouTube Still Learning — Or Just Repetition?

When content production cost collapses, volume rises but novelty compresses.

№054 · Aug 16, 2026

Why Are We Still Communicating Like Video Is Expensive?

The cost of producing visual explanations has collapsed by 90-99% with tools like Sora, Veo, and Synthesia.

№053 · Aug 14, 2026

Stop Ruining Email

AI has driven the marginal cost of sending email to near zero, triggering a self-reinforcing arms race: volume-driven outreach floods the channel, aggressive filtering catches…

№052 · Aug 11, 2026

Compute Is Not Free Lunch

Don't ask what AI costs — ask what your people can do with it.

№051 · Aug 8, 2026

The Next Big AI Market May Be Boring Services

The biggest AI opportunity of the next decade may not be building frontier models or flashy AI products.

№050 · Aug 5, 2026

The Best AI Opportunities May Not Look Like AI Companies

Riding the AI wave does not require building an AI software company.

№049 · Aug 3, 2026

AI strategy · AI transformation 14 min

The Fixed Pie Fallacy

When AI collapses execution cost by 10x, the rational response is not to do the same work with fewer people — it is to do dramatically more work.

№048 · Jul 31, 2026

When Intelligence Becomes Cheap, What Becomes Expensive?

AI is driving the marginal cost of intelligence toward zero, but abundance does not eliminate scarcity — it migrates it.

№047 · Jul 28, 2026

AI strategy · AI transformation 14 min

Enterprise AI Will Be Slower Than the Demos

Enterprise AI is not slow because the models can't do the work.

№046 · Jul 25, 2026

AI strategy · AI transformation 8 min

Why AI Pilots Fail (And Why You Should Run Them Anyway)

AI pilots can build organizational skill even when they produce no immediate return. The test is what the organization learns and retains.

№045 · Jul 23, 2026

Strategy Looks Different When You Have Had to Sell It

AI has made strategy generation nearly costless while the ability to evaluate options against execution reality has not scaled.

№044 · Jul 20, 2026

What We Learned Before AI — And Why It Matters Again

The operational disciplines that taught organizations to decompose, analyze, and improve systems did not become obsolete — they became dormant.

№043 · Jul 17, 2026

Your Job Was Never the Job

Between 57 and 60 percent of knowledge work is not the work — it is the translation layer that exists because the execution layer is human.

№042 · Jul 14, 2026

The End-to-End Operator: Why the Most Valuable People See the Whole System

As AI cheapens execution and amplifies abstraction, the scarce skill is end-to-end visibility — the cognitive habit of moving between abstraction layers, diagnosing root causes…

№041 · Jul 12, 2026

organizations · AI transformation 13 min

The Business-IT Divide Was a Workaround. AI Is Removing What Made It Necessary.

The separation of business and IT was never an organizational preference — it was a rational workaround for a complexity problem.

№040 · Jul 9, 2026

Why Most No-Code Tools Are a Dead End

No-code tools abstracted syntax but not structure.

№039 · Jul 7, 2026

AI models · AI strategy · architecture 11 min

Frontier Models Are Becoming Operating Models

Frontier models are becoming operating models for cognition, not just tools — and that changes the strategic question

№038 · Jul 5, 2026

AI models · memory · AI architecture 11 min

Long Context Is Not the Same as Better Memory

Long context and memory solve different problems; confusing them is the most common error in AI system design

№037 · Jul 3, 2026

AI models · org design · strategy 11 min

One Brain or a Team of Specialists?

Frontier models embody different architectural philosophies — model choice is becoming an org-design decision

№036 · Jul 1, 2026

AI models · reasoning · how AI works 12 min

Why 'Just Pattern Matching' No Longer Explains What These Models Do

A precise framing of what these models actually do — beyond both the hype and the dismissal of 'just autocomplete'

№035 · Jun 28, 2026

organizations · AI transformation · management 10 min

The Org Chart After AI: Fewer Layers, More Friction

AI cheapens coordination but makes alignment more expensive — flattening without redesign creates instability

№034 · Jun 23, 2026

AI context · automation · knowledge management 10 min

Curate Before You Automate

Treat curated external knowledge as owned infrastructure; curate the reference layer before you automate against it

№033 · Jun 18, 2026

AI adoption · strategy · organizations 12 min

The Dissipation Gap

The real AI advantage is not the technology — it is the distance between what is possible and what your organization can absorb

№032 · Jun 13, 2026

metrics · AI transformation · organizations 12 min

You Are Measuring the Wrong Things in the AI Era

When output is no longer the constraint, the metrics that built your organization become the thing holding it back

№031 · Jun 8, 2026

AI strategy · vendor lock-in · infrastructure 12 min

Don't Build Your AI Strategy on Someone Else's Stack

A framework for AI infrastructure risk, grounded in what frontier companies themselves disclose about dependency

№030 · Jun 3, 2026

creativity · AI strategy · taste 11 min

When Creation Becomes Abundant, Taste Becomes Infrastructure

As AI makes creation abundant, the scarce human contribution shifts to taste: knowing what should exist and being accountable for it

№029 · May 28, 2026

AI infrastructure · strategy · execution 11 min

The Brain in a Jar

Choosing an AI harness is not a tool decision — it is an organizational philosophy decision about how humans and machines divide work

№028 · May 25, 2026

geopolitics · cost structure · ai operations 12 min

The Coming Cost Shock: When China Combines Cheap Labor with Cheap Intelligence

AI is shifting from a productivity tool to a cost-structure determinant. As China converges cheap labor, cheap energy, and cheap intelligence into full-stack cost compression, incremental adoption won't be an answer.

№027 · May 22, 2026

neural symbolic · hybrid architecture · system design 9 min

The Future Is Not Neural or Symbolic. It Is Both.

The neural-vs-symbolic debate is a false choice. The most effective AI systems are hybrid — neural for ambiguity and synthesis, symbolic for exactness and verification.

№026 · May 19, 2026

physical work · iteration · digital twins 13 min

The Iteration Gap

AI's biggest gift to physical businesses isn't automation — it's the ability to iterate a thousand times digitally before committing once in the real world. The ones who figure this out first will outcompete the ones who don't.

№025 · May 18, 2026

AI strategy · AI transformation 14 min

The New Bottleneck Is Not Intelligence. It Is Accountability.

AI has made intelligence cheap.

№024 · May 17, 2026

trust-architecture · zero-trust · structural-safety 12 min

Trust Is Not an Instruction

A framework for redesigning human-AI trust from behavioral compliance to structural controls -- and why the difference determines whether AI deployments remain safe at scale

№023 · May 15, 2026

Forward Deployed Engineers: The Fastest Path to Products That Work

The Forward Deployed Engineer model — embedding your best technical people inside a customer's actual workflow to discover what truly needs to be built — is not a Palantir trick.

№022 · May 10, 2026

systems-thinking · system-design · operational-excellence 7 min

The System Is the Product

A framework for understanding why improving outcomes requires improving the system that generates them, and why AI makes system design the dominant factor in performance

№021 · May 8, 2026

productivity · mental-health · ai-work 11 min

The Anxiety of Infinite Capacity

AI removed the execution bottleneck. What rushed in was not a golden age of productivity but a quiet, persistent anxiety the old constraint had been holding at bay.

№020 · May 6, 2026

AI agents · productivity 11 min

The Air Traffic Controller Syndrome

One hour with AI is not one hour anymore.

№019 · May 3, 2026

AI agents · productivity 15 min

The Overnight Shift You Never Clocked Into

Agent orchestration imposes air-traffic-control-level cognitive load on knowledge workers — parallel monitoring, rapid context switching, vigilance maintenance — but without…

№018 · Apr 24, 2026

branding · ai-disruption · agentic-commerce 11 min

The Brand Crisis: When AI Decides What You Buy

AI agents are redirecting trillions in consumer spend. Brand equity built in human minds does not automatically transfer to the algorithm making purchase decisions.

№017 · Apr 12, 2026

What AI Actually Changes in Large Organizations — and What It Does Not

AI compresses the cognitive layer of organizations — the phases, layers, and coordination that existed to manage human limitations — while leaving politics, physics, and…

№016 · Apr 10, 2026

decision-making · ai-productivity · organizational-design 9 min

Faster Execution, Worse Decisions

AI eliminates execution as the organizational bottleneck. Decision architecture was not redesigned for it. The result is organizations moving faster in directions nobody validated.

№015 · Apr 9, 2026

Why AI Initiatives Fail — And It Is Not the Technology

Most AI initiatives fail not because the technology underperforms, but because it is applied to systems that were already misaligned.

№014 · Apr 6, 2026

The Platform Ate My SaaS

The SaaS middleware layer is being absorbed by model providers and hyperscalers spending $690B to own downstream token consumption.

№013 · Apr 5, 2026

Four Skills for the AI Era

The capabilities that matter most in an AI-saturated job are not technical.

№012 · Apr 3, 2026

It Used to Be Simple to Build. Now It Is Again.

Building software follows a cyclical accessibility pattern — transparent but limited, then powerful but impenetrable, now powerful and accessible again.

№011 · Apr 2, 2026

The Cheap Experiment

The cost of experimentation just collapsed to near-zero — and the organizations that recognize this will run thousands of experiments while competitors are still planning their first.

№010 · Mar 25, 2026

saas · ai transformation · organizations 8 min

The SaaS Collapse

A framework for understanding which SaaS products survive AI platform commoditization and why domain knowledge is the only durable moat

№009 · Mar 21, 2026

antitrust · regulation · algorithmic pricing 13 min

No Meeting Required: How AI Broke Antitrust Without Breaking the Law

AI pricing algorithms achieve outcomes indistinguishable from collusion — without any agreement between competitors. The law wasn't built for this.

№008 · Mar 17, 2026

project management · organizations · ai transformation 8 min

Why Project Management Is About to Look Very Different

AI is compressing the cognitive scaffolding of project management. The phases that remain are the ones that serve physics, politics, and accountability — not information processing.

№007 · Mar 10, 2026

ai adoption · workplace · talent 9 min

How Many Tokens Come With My Job?

AI access is following the same arc as mobile phones, laptops, and internet. The companies that understand this are hiring from the companies that don't.

№006 · Mar 3, 2026

ai adoption · data · labor 10 min

The Human API: Why AI's Next Bottleneck Is You

AI is running out of human-generated data. The stock of high-quality public text will be exhausted by 2032. The machines need us more than we think.

№005 · Feb 26, 2026

AI strategy · AI transformation 9 min

Why I Stopped Using ChatGPT

I did not stop using AI — I stopped using it through a chat window.

№004 · Feb 24, 2026

knowledge management · organizations · ai transformation 12 min

Fine-Tune Your Organization

Your company spent millions on knowledge management. The return was a graveyard with excellent uptime. Here's what organizational fine-tuning looks like instead.

№003 · Feb 17, 2026

expertise · talent pipeline · knowledge transfer 9 min

The Expertise Cliff: Who Judges When Nobody Learned?

The top of the talent pipeline is retiring. The bottom is closing. What happens when the people who carry knowledge leave and nobody was there to absorb it?

№002 · Feb 10, 2026

education · ai adoption · knowledge transfer 10 min

The Teacher's Last Monopoly

AI tutoring doubled learning gains at Harvard. Duolingo created 150 courses in one year. The knowledge transmission monopoly ended — what remains is the power to design how AI teaches.

№001 · Feb 3, 2026