🟡 5 MINUTE SUNDAY · Every Sunday. One tip. Under 5 minutes.
Last week I stood in a room in Berlin full of Chiefs of Staff. People who run the company from the inside. Almost none of them are in the org chart as the reason it works. I gave them one project: run an AI transformation, start to finish, in 7 days. Not a pilot. Not a proposal deck. A real change, with a real ally, measured by Friday.
Most people in that room thought I was joking.
Most AI rollouts don't fail in the first month. They fail in the first meeting. The one where you pitched the feature instead of the fear. And deep down, you already know that.
— 01 — ONE LEADERSHIP TIP
McKinsey counted it in November.
88% of companies use AI somewhere in the business.
Only 6% see it move the P&L.
MIT ran their own count and landed close by, most pilots that die never had a technology problem. They had an ownership problem. Gartner backs it up too, 30% of generative AI projects get abandoned after the proof of concept, most for the same reason. Nobody owned the room.
I built a framework to fix that, and I ran the whole thing live with the Chiefs of Staff last week. Here's exactly what we did.
Exercise 1: The Stakeholder Reality Check.
Before anyone touches the tool, they map six people. CFO. Head of Engineering or IT. HR. The team who'll actually use it daily. The CEO. And the one who could kill it quietly, not your enemy, just someone who's never been asked what they're afraid of.
For each one, three questions. What do they actually care about, not what you think they should care about. What makes them say yes. What makes them say no.
Then two names, out loud, in front of the table. Who wins most if this works, that's your ally, you talk to them first. And who could kill it quietly, that's the conversation you have before you pitch anyone else.
Exercise 2: The Translation Game.
The rule is simple. If you can't say why this matters in one sentence that serves their goal, you haven't translated it yet. Features describe the tool. Translation speaks to their problem.
I gave them a real example, an AI meeting notes tool. To a CFO: saves 8 hours per manager per month, at your salary band that's real money back into productive time. To a Head of Engineering: runs in your existing stack, data stays in your tenant, no new vendor to manage. To HR: managers spend 30% less time on admin, that time goes back to their teams. To the CEO: nothing falls through, every meeting ends with a clean action list, automatically.
Same tool. Four completely different pitches. None of them mention the AI.
Exercise 3: The 7-Day Experiment.
This is the part that made the room go quiet. Write down, in one sentence, the problem the initiative solves. Not the AI problem, the business pain underneath it. Name your ally by first name, not the team. Name the concern you'll address first, what that person is actually afraid of before they say it out loud.
Then the smallest possible action. One person, one metric, seven days. Nothing that needs training. See who uses it first publicly, the most credible person you've got. Try it with the smallest group, zero friction. Share the win somewhere visible, Slack, an all hands, a 1:1 with the CEO. If the win doesn't get shared, adoption stops at the pilot. Every time.
The Metrics Card.
Thirty days after the pilot, come back with four numbers. Activation rate: who used it at least once in week one, target 100% of the named individuals, not the team. Return rate: who used it again without being asked, weeks two through four, 60%+ is strong adoption, below 30% means the use case is wrong, not the tool.
One business metric, picked in advance: time to first draft, meetings turned into sent actions, or first response time on support tickets. And acceptance rate at week two, the share of AI outputs people actually kept without rewriting, 60%+ means it's the right fit.
Google's own research puts the ceiling at 105 minutes saved per person per week, with a 70% acceptance rate on suggestions when it's built right. Worklytics found 60%+ return rate is the healthy benchmark after 30 days. And McKinsey's number is the one that should worry you: only 6% of organizations using AI see it show up in the financials at all.
The sentence you're building toward, the one that gets you funded for phase two: we went from X to Y on this task in this many days, that's this many hours back per person per month. If you can't fill in that sentence, phase two doesn't get funded.
The takeaway: you don't need a transformation roadmap. You need one person, one metric, and a week.
— 02 — ONE COMPANY WORTH KNOWING
Harvey, the legal AI startup that sells to lawyers, not to IT departments.
They don't launch with a company wide rollout. They start with the partners who already trust the firm's leadership, get one visible win on one real case, then let the associates ask for access themselves. Adoption follows credibility, not a launch email. It's the same sequence I taught in Berlin, they just run it at company scale.
The lesson: find your one ally before you build the rollout plan.
— 03 — ONE THING TO ACT ON
Do this before next Sunday.
- Pick one AI initiative you're stuck on. Not the biggest one, the one that's actually stalled.
- Map three stakeholders. What they care about, what makes them say yes, what makes them say no.
- Name your ally, one name. Name the person who could quietly kill it, and book five minutes with them first.
- Give one person one action for seven days. Write down the number before. Write down the number after.
⏱ Time required: 25 minutes.
This is the exact framework, worksheets included, that I ran live with 40 Chiefs of Staff last week. DM me AUDIT and I'll send you the whole thing.
See you next Sunday.
— Alisa
Alisa Reznik · Fractional COO · Executive Advisor · Keynote Speaker
I help founders scale the business without losing themselves in it.
This issue also ran on LinkedIn — join the discussion there
