A brief that writes itself. A P&L you can question. A red team for your next big call.

Set up once, pays you back every week: your morning brief, your P&L, and your biggest decision.

So, what’s DeskTheory all about??

I don’t know about you, but my head is spinning at an increasing rate as we experience the hard takeoff in AI. Every day there’s a month’s worth of news.

I started DeskTheory because the noise was driving me nuts, and as a founder of 4 businesses I needed to know where I could get real leverage today in my role as CEO personally and in my businesses.

I became obsessed with learning about what’s really happening with AI and how I can use it to massively improve my businesses. Then my CEO friends started booking countless calls with me to get me to teach them about how I was using AI.

That led to the first two books I wrote about AI for CEOs. They were based on hundreds of hours of conversations with CEOs (I transcribed them all and use systems shared in DeskTheory to create the book outlines). I started sharing them with my friends so I could avoid having the same conversation over and over with every CEO I know.

Then I had an idea. Why don’t I create the high signal AI information source I wish I had and that I’d share with my friends? That became DeskTheory.

My mission is twofold:

1. To take each of you from wherever you’re starting from to among the top 1% of CEOs using AI (and by extension, absolutely dominate in your industry)
2. Be clear signal in a sea of noise about AI

I hope you enjoy DeskTheory. Send me a note if you have any feedback or want to learn about anything in particular.

Andrew

1 · A daily brief, written while you sleep, ready when you wake up

The outcome. A one-page brief waiting for you every morning, assembled before you wake: what came in overnight that needs you, today's calendar with the meetings to prep flagged, the one number you watch, and what's sitting on you to decide. You start the day deciding instead of spending your freshest hour reconstructing where things stand across five tabs.

The setup. A scheduled Claude Code routine that runs on Anthropic's servers, even with your laptop closed, reads your connected inbox and calendar, pulls one metric, and writes the page at 5am. Get the prompt right by hand for a few mornings first, then set it on a daily trigger.

What you should do next. Tomorrow morning, build the brief as a one-off prompt and run it. Once it's the page you actually want, schedule it as a cloud routine. From then on you read a brief instead of building one.

Full step-by-step → Read the full workflow

2 · Talk to your P&L in plain English

The outcome. Upload your profit-and-loss once and interrogate it like a person. Why did gross margin drop in March? Which costs grew faster than revenue? You get a grounded answer with the math shown, in seconds, instead of waiting days for a custom pull by the time the moment to act has passed. The questions you never asked cost more than the ones that came back late.

The setup. A Claude Project on a business plan, your P&L exported as a trailing-twelve-month file, and three facts in the instructions: your accounting basis, which rows are subtotals, and "only use numbers from the file." That last setup detail is what keeps the answers honest instead of confidently wrong.

What you should do next. Export your last twelve months and brief the Project. Ask the question you'd normally park for your next finance sync, then verify one figure against the source statement. Watch the lag from "I wonder why" to "now I know" collapse from days to seconds.

Full step-by-step → Read the full workflow

3 · An AI red team for your next big call

The outcome. The people around you are not incentivized to tell you a big decision is a bad idea. The deal team wants the deal; nobody wants to be the one who slows the room down. An AI has no ego and no career risk, so it asks the uncomfortable question before you wire the money.

The setup. Write a one-page memo on the decision, then explicitly assign the adversary role: "You are my red team. Find why this fails. Run a premortem, argue the strongest case against, and name the three assumptions that, if wrong, sink it." Assigning the role is the whole trick. Ask it nicely and you just get a smarter cheerleader.

What you should do next. Take the biggest decision on your desk this week and run it. Treat the output as a checklist, not a verdict, then make the call with the holes visible. Run it on the decision you feel most sure about. That is exactly where your blind spots are biggest.

Full step-by-step → Read the full workflow

P.S. Card 1 and card 3 reward the same habit: get the prompt right by hand once, then let it run. Card 2 you will be using ten times a week by Friday.