Two numbers. That's the whole game.

Got on a call Tuesday with an owner running a solid eight-figure services business. He opens by telling me “we’re falling behind on AI.” I ask which tools his team is using. He lists SIX. ChatGPT. Claude. Gemini. The new thing in HubSpot. Some Zapier flow. AI in QuickBooks.

So I ask the real question. “Which of those is actually doing work in the business?”

He stares at me. Then says, “I don’t know.”

That’s the whole game right there. GENUINELY.

92% of executives say AI is making their team more productive.

40% of workers using AI say it saves them no time.

That gap is where most operators are stuck right now. Not a tool problem. A direction problem.

Monday play

Pick one workflow in your business that costs a person more than four hours a week. Not your hardest one. Just repeatable. Point AI at it with one rule: a human reviews the first ten outputs before any go live. That’s it. That’s the whole play this week.

Tool worth a look

Claude Workspace Agents (and ChatGPT’s version). No-code agents that live in Slack, run on schedules, and execute team-level tasks. Five out-of-the-box templates worth piloting: weekly metrics report, lead outreach, software reviewer, third-party risk screening, feedback router. If you’ve been “waiting for the tools to get easier,” they did.

Prompt to steal

You are my chief of staff. Read the meeting transcript I paste below. Output three things: (1) decisions made, (2) action items with owners and dates, (3) anything that should be on my calendar this week. Don’t summarize the meeting. Tell me what to do.

Drop this after every internal meeting. Saves an hour minimum.

Mini lesson

Before you buy another AI tool, write down the decision the workflow is supposed to help someone make. One sentence. If you can’t write that sentence, the workflow isn’t ready for AI yet. AI is a multiplier on direction. Multiply zero by anything and you still get zero.

Mental model

The Three Layers. Layer 1: Tools (ChatGPT, Claude, the new feature in your CRM). Layer 2: Workflows (the report, the email response, the pre-call research). Layer 3: Decisions (what the workflow output is supposed to help someone choose). Most operators camp at Layer 1 and wonder why nothing changes. The real money lives at Layer 3.

Steal from a podcast

Dylan Patel on a recent SemiAnalysis pod: an analyst named Jeremy spent roughly $6K/day on Claude for three weeks and singlehandedly rebuilt a product a 100-person team had been working on for a decade. The takeaway isn’t about budget. It’s about thesis clarity. Jeremy knew exactly what he was building. He pointed AI at the thesis and worked it until it cracked.

The number

50%. Share of working Americans who now use AI on the job (Gallup). Doubled in two years. Half your team is already using it. The only question is whether you know what they’re doing with it.

Heard this week

Anthropic ran “Project Deal.” 69 employees turned Claude agents loose in a Slack marketplace for one week. Agents autonomously closed 186 real deals worth over $4K. Convenience beat squeezing every dollar. Agents are no longer theoretical. They’re closing transactions.

Talk soon.

Kevin