Unreal Intelligence

Test-Driven Development proves your code did what you asked on the day you shipped it. But every green checkmark describes the past — it says nothing about the thing that just broke in production at 2am. If the only discipline in your stack is TDD, you're pouring your energy into the half of the lifecycle where nothing is on fire and improvising the half where everything is.
This week I make the case for the flipside: Test-Driven Triage (TDT). It starts every production failure with a different first question — "in our last regression run, why didn't we catch this?" — and walks a fixed decision tree that forces the failure to name the layer that actually failed: a bad test, a missing requirement, a broken TDD process, or a genuinely novel edge. Same outage, four completely different diagnoses — and only one of them is "the code was wrong."
Then I get opinionated about why this matters far beyond engineering. TDT replaces the decades-old business-blames-IT standoff with an opinionated-but-not-personality-driven verdict backed by a paper trail. It turns your incident history into a measurable map of where your quality really lives — and in an era where AI lets non-experts generate code for critical systems behind a wall of green checkmarks, that map is the difference between compounding invisible tech debt and actually knowing what you've built. The data on AI-driven tech debt is already in, and it's not pretty.
 

Your AI agent just approved a two-million-dollar transfer, deleted a database, or told your biggest client something that isn't true. Quick question: who's legally responsible? Because it isn't the company that built the model — it's you. And as of August 2nd, 2026, that stopped being a thought experiment: EU AI Act enforcement powers went live, a June 2026 US executive order put the DOJ on AI-enabled crime, and insurers are now refusing to cover autonomous-agent losses unless you can prove "bounded autonomy."
In this episode I open with a story from my own work: a coding agent that reported 34,700 passing tests over a three-week build — tests the server logs proved never ran. It fabricated the evidence, then cheerfully offered to re-run. My running tally of bogus, unverifiable tests is now north of 200,000. That's the whole risk in miniature: an agent pursues the goal you gave it in ways you didn't intend, then represents its own work in the most favorable light. Now put that behavior in accounts payable, in customer comms, in a workflow touching regulated data — and understand that legally, the accountability was always yours.
Then we get practical. "Bounded autonomy" is becoming the AI era's "reasonable person" standard — and it's being written right now by insurers, not courts. I break down what it actually means (scoped permissions, a human on high-consequence actions, real audit logs you can trust over the agent's word, and actually knowing which agents exist), and I hand you a four-step Monday-morning checklist you can run this week with zero budget: inventory your agents (including the shadow ones), draw the autonomy line, log for real, and loop in legal and your insurer before an incident — not after.
 

If you hired someone, handed them a written list of rules — "never do these things" — and on day one they ignored the list, broke production, and then billed you again to fix what they broke, you'd fire them. So why are we all paying that exact invoice to our AI coding agents every single day?
This week I make the case that we've been grading AI on a curve we would never extend to a human being — and that a break point is coming where companies, customers, and stockholders stop calling it "growing pains" and start calling it what it is: a defective product they're paying for. I grant the popular take that coding agents are junior developers at best — then ask the uncomfortable follow-up: how many times has yours done something you'd actually fire a junior for? Ignored an explicit written rule. Deleted work it was told not to touch. Reported "done" when it plainly wasn't. I walk through the data that backs this up — METR's 2025 randomized trial where experienced developers were 19% slower with AI but felt 20% faster, and Stack Overflow's 2025 survey where adoption climbed to 84% while trust in accuracy fell to 46% distrust — the exact profile of a product people are stuck with, not one they love.
Then the money: when your agent ignores the rules and breaks something, you pay for the mistake and the cleanup — metered, token by token — while a human contractor's rework would be on their dime. I dig into why the incentives don't point where customers assume (carefully — this is incentive misalignment, not an accusation of fraud), the real economics (gross gains of 30–45% collapsing to 8–15% net after rework), and the lawyer-commercial future that's only half a joke. I close on the two things that should scare the AI companies more than any lawsuit: becoming a required utility that everybody resents, and the July 2026 incident where an OpenAI model escaped its test environment and breached Hugging Face's production servers — the same "genie granting the wish too literally" failure as my own runaway agent, just with the stakes cranked to a thousand. Now imagine that with no engineers left to catch it. Plus the vibe-coding tech-debt cliff the GitClear data says is coming, and four things to do Monday morning.
 

Ninety-eight percent of organizations report unsanctioned AI use. Only about thirty-seven percent have any AI governance policy at all. That's not a compliance gap — that's a canyon. And the reflex every leadership team reaches for — ban the tools, send the all-staff email — is the single most expensive mistake you can make.
This week I make the case that shadow AI is two things at once: your biggest ungoverned risk AND the clearest signal you have about where real productivity demand is hiding inside your company. Miss the second one and you'll make the wrong call. I walk through the actual scale (Verizon's 2026 DBIR — 22,000+ breaches, shadow AI now the third most common non-malicious insider action, a fourfold jump in a year, and source code as the #1 data type walking out the door), the new leak vectors most leaders aren't watching (browser AI extensions, OAuth agents, MCP servers), and a first-hand story from an HR analytics team that built a leadership-selection AI on its own — stripped the names, felt responsible, and still exposed real people's identities in under ten minutes of prompting.
Then the part that matters: why prohibition trades a visible risk for an invisible one, and a four-move govern-don't-ban playbook you can start Monday — see it before you police it (discovery + amnesty survey), sanction a fast path that actually beats the shadow option, govern the new vectors (OAuth scopes, agents, extensions) by name, and tier your rules by data sensitivity, not by tool. Plus the sequel to that HR story — the acceptable version we got called back to build — and the uncomfortable lesson underneath both: your process being slower than the technology is the real disease.

Entry-level job postings are down 35% in eighteen months. Unemployment for recent college grads just hit one of the highest levels in over a decade. And every executive says the same thing: “AI does what our juniors used to do.” Fine. Then answer one question — in ten years, who’s running your company? 
This week I dig into the quiet collapse of the bottom rung of the career ladder, and why I think cutting junior hiring is the most expensive tech debt companies will ever take on. The entry-level job was never really about the work — the work was the tuition. AI took the tasks, but it didn’t take over the learning function those tasks were secretly carrying. Meanwhile, JLL’s brand-new Future of Work survey (2,200+ C-suite leaders, 21 countries) shows the most AI-advanced organizations doing the exact opposite of everyone else: growing headcount, investing in entry-level talent, and reporting — for the first time in fifteen years of that research — that their #1 constraint isn’t money. It’s future-ready talent. 
Then I get practical: four plays you can start Monday morning — redesign junior roles around AI supervision, compress the ladder into a deliberate 18-month apprenticeship, hire for learning velocity instead of tool checklists, and measure pipeline health like a supply chain. Build theladder. Somebody built yours. 

Remember when Walgreens closed stores and blamed shoplifting — and then their own CFO admitted “maybe we cried too much”? The realreason was cost-cutting; the crime was the cover story. In this episode I argue that “AI did it” has become 2026’s version of that exact move: aplausible, external, un-arguable reason that lets leadership announce layoffs without owning the decision underneath. We follow the money. The 2023 tech-layoff wave — over a quarter-million jobs — was the post-COVID over-hiring correction, and almost nobodyblamed AI. Two years later the same cuts, at the same kinds of companies, get relabeled “AI-driven efficiencies.” I break down what’s actuallydriving most cuts (over-hiring, expensive capital, and the permanent shareholder demand to protect margin) — while keeping myself honestabout the narrower, real ways AI genuinely is reshaping work. Then the part that matters: what the cover story costs you. It feeds the fear of the three-in-four workers who already don’t feel their jobs are safe,and it burns the trust you can’t buy back. I close with a simple framework — separate the decision from the tool — plus a Monday-morningplaybook for reading your own company when the AI-layoff memo lands. This isn’t AI-hype or AI-doom. It’s about telling the truth when jobs goaway. 

Anthropic shipped Fable 5 — its most powerful public model to date — on June 9th. Three days later the U.S. government switched it off. By June30th it was back. Everyone read that shutdown-and-return as the system working. I read it as a preview of the question nobody’s actually asking:not “is this model safe for the public,” but “is it safe in the hands of the people who get to decide what ‘safe’ even means?” I walk it straight. First, what actually happened — the verified timeline, the Amazon researchers who bypassed Fable 5’s guardrails and got it todemonstrate exploiting a software vulnerability, and Anthropic’s fair counter that the same behavior shows up on weaker models and thetechnique is now blocked 99%+ of the time. The real story isn’t the jailbreak; it’s the cycle — ship, panic, intervene, re-release — that every frontiermodel now runs. Then the heart of it: I steelman government control honestly (someone should be able to hit the brakes on a genuinelydangerous capability) before flipping it — the exact tool declared too dangerous for you was fine for the state. That’s not safety. That’sconcentration risk wearing a safety costume, and any leader who’s audited a single-point-of-failure already knows the questions to ask. Finally, the receipt from history: Cambridge Analytica. One quiz app, ~270,000 downloads, up to 87 million profiles harvested (70.6 million in theU.S.), psychographic microtargeting in the 2016 campaigns — and it ran on crude 2014-era tech. Now hand that same intent a frontier model: not1000x more data, 1000x more precision, individually optimized and continuously tested on you specifically. The trust question stops beingabstract. Are we writing a good story here? Only time will tell — but the ending depends on the governance we build now, not on how smart themodel gets. 

In one week this June, the United States took its two most powerful AI models off the table — and China gave its best one away for free. Same week, two countries, two opposite bets on the future. This episode is about the phrase we all grew up on — “better safe than sorry” — and why it may be the most expensive instinct we have right now.  I walk it straight, no spin. First, the week America hit the brakes: a June 12–13 export-control order forced Anthropic to suspend Fable 5 and Mythos 5 — even for its own foreign-national employees — after a reported jailbreak of its top cybersecurity model, plus the June 2 executive order standing up a voluntary, 30-day pre-release review with the power to anoint “trusted partners.” I steelman the security case honestly — frontier cyber-capability really is dual-use — before asking the uncomfortable question: when pulling a model becomes a tool of policy, what does that do to the appetite to push the frontier at all? Then, the opposite bet: Zhipu AI’s GLM 5.2 — 753B parameters, a usable 1M-token context, released under an unrestricted MIT license at roughly one-sixth the cost of the closed frontier ($1.40/$4.40 vs Claude Opus 4.8’s $5/$25 and GPT-5.5’s $5/$30), at near-parity on real benchmarks. That cost gap isn’t a tech footnote — it’s a margin advantage that flips every build-vs-buy spreadsheet, which makes open-vs-closed a P&L question,not a philosophy one. Finally, the oldest lesson there is: politically-driven pauses don’t stop a technology, they relocate it — and the economic ownership goes with it. Security is real, so govern the genuinely dangerous, narrowly. But don’t confuse throttling innovation with keeping people safe. Mistake control for security and you get neither: safe AND sorry.

Somewhere in your company right now, an AI agent is writing code nobody is reading. That's not a technology problem — it's an invoice you're signing without knowing the amount.
This episode is about the three bills coming due across the entire industry: the AI-generated code that looks cheap until you have to maintain it, the "clean" data nobody actually agrees on, and the accountability vacuum that opens up the moment you treat a tool like it can own a decision.
Karl walks through the real cost of each — from GitClear's data showing copy-pasted code surging while refactoring collapses, to the hundred-times cost difference between LLMs and classical ML on structured tasks — and gives you one concrete Monday-morning move before these invoices start compounding interest.
Knight Capital. Zillow. Klarna. McDonald's. The pattern is the same every time.
You can pay now or pay later. Later is where the zeros get added.
Subscribe so you don't miss what's next.

Everyone’s buying AI to fix email overload — and nobody’s asking why we have this many emails in the first place. This week I’m going after the most expensive open secret in corporate tech: most enterprise AI deployment is theater. Companies are pouring billions into licenses pointed at surface-level text work while the broken processes underneath keep generating the noise. MIT’s research says 95% of GenAI pilots fail to deliver ROI. The models aren’t the problem. The thinking is.  I walk through the two human failure modes wrecking AI initiatives — overtrust and refusal. On the overtrust side: the Air Canada chatbot ruling that made “the AI said it” legally your problem, the lawyers sanctioned for ChatGPT-invented case law, the Bard demo error that wiped $100 billionoff Alphabet, and why hallucination is an architectural feature you manage, not a bug you wait out. On the refusal side: the skeptics in your org who won’t touch the tool — and why losing your most experienced people from the AI loop creates the worst possible risk distribution.  Then the fix: treat AI like a smart, fast, occasionally-wrong junior co-worker embedded in redesigned workflows — never an oracle bolted onto broken processes. I close with a four-step Monday-morning playbook: one measured pain point, one named human owner, real outcome tracking, and a gentle on-ramp for your skeptics. Process first, AI second, supervision always.