God Mode is breaking work

Hiya 👋

There's a shift happening inside companies adopting AI that most leaders are feeling but haven't named yet. It's quieter than the headlines, and it's the thread running under this whole edition. The cost panic, the new roles showing up, and the playbook at the end are all the operator's side of it. Chapters of the same book.

Let's get into it 👇

Growing up, I used to type "cheese steak jimmy's" into Age of Empires. It handed you every resource in the game, instantly. Some people called it God Mode.

It was fun for about ten minutes, and then the game was boring, because the challenge was the entire reason to play.

I keep thinking about that watching teams adopt AI. Everyone's arguing about jobs: which roles are safe, when the wave hits which industry. The thing I'm seeing firsthand is quieter: People who spent 5, 10, 20 years getting good at the hard part, the problem nobody else could crack, are watching that part get done for them. And the hard part is where their meaning lived.

Most people aren't saying this out loud, which is exactly why it's worth naming. It splits into two problems leaders are holding at once, while they're still separable.

• The operational one: the window to build and own the workflows your business runs on is open, and getting easier to do.

• The human one: the purpose conversation is coming to your team whether you're ready or not, showing up as quiet disengagement long before it's anything you can name.

If you’re a leader at a company, the question worth taking into your next leadership meeting is this: where does the purpose come from in what we do and how we do it in a world where AI can help our team do a lot of their work?

I wrote about this in a lot more depth below.

Go deeper:

🎮 Read the full piece → God Mode is Breaking Work

While the above is about the shift in the human side of business, Derek Thompson says, incorrectly, that the AI shift has entered its "wait, is this worth it?" phase.

What's the story? An AI consultant told Axios that one client spent half a billion dollars on Claude in a single month. Ramp's lead economist says the average business now spends 13x more on AI tokens than in January 2025. Uber and Microsoft have reportedly walked back their “use AI as much as possible” tokenmaxxing mandates.

There’s two things actually happening here:

1. The panic has less to do with the technology than with how companies told people to use it. As execs pushed everyone to "use AI more" and usage itself became the metric, some firms literally ranked employees by token consumption. Classic lesson of “just because you can, doesn’t mean you should.”

2. Almost all of these cases are engineering teams who combined “tokenmaxxing” with huge amounts of work being done with the highest end models with the highest reasoning capabilities used.

Why anyone is surprised that the Ferrari being driven around the track at 200km/h is burning more fuel than the Toyota driven to drop the kids off at school is beyond me.

Test it for yourself: if you’re not a huge AI user, check your usage stats (Claude version, OpenAI ask your admin) and then use a standard model and throw 5 PDFs or large spreadsheets at AI and ask it to analyze it in a way that’s useful to you. I would bet you that your usage needle barely moves. This is how most people are using AI right now and not at all like engineering teams are using it who derive a ton of value but, as engineers tend to do, over-engineer the system for performance over efficiency.

All this is to say, the “fog of war” around AI stories is high but some critical thinking beyond the clickbait headlines shows it’s a lot of hot air.

Go deeper:

💸 Read the panic → The Great AI Cost Panic of 2026

The New York Times shares a great case study that doubles as a preview of the org design questions headed your way.

What's the story? Box started weaving AI into its products a few years ago and has created 13 entirely new types of jobs: AI architects, AI solutions managers, AI platform leaders.

The most telling one is a senior director of AI, data and integration, hired to wire Box's internal systems and data together so people could finally use AI in their work. Box expects to pass 3,000 employees this year while much of its industry cuts.

Strip away the Silicon Valley scale and the titles collapse into three functions every operating company will need someone to own, whether or not they ever become full-time jobs:

1. Own the plumbing: Box's integration director exists because AI is only as useful as the systems and data it can reach. At 150 people this is a mandate rather than a hire: one named person responsible for how data moves between your CRM, your ERP, and the spreadsheets in between.

2. Spread what works: Box's "AI business automation engineers" sit in IT and help colleagues get repetitive work off their plates. You almost certainly have this person already, working informally in stolen hours. Give them the mandate and a few protected hours a week.

3. Decide what's good enough: Box hired people specifically to evaluate AI models and outputs. At your scale this is a standard: who approves an AI-touched workflow before it runs on real customers, and against what bar.

If you run a small or mid-market company: these functions are starting to land in others org charts. Planned, they're three wins for the companies that are seizing this new era. Unplanned, they're the gaps that surface later as stalled pilots burned-out informal hero’s trying to make AI work in a company not meeting the times.

Go deeper:

📦 Meet the new roles → How Box Created 13 New Types of Jobs Because of A.I.

Know someone who’d enjoy this newsletter?
Forward it along, or send us a note if there’s a topic you’d like us to unpack next.