“One either meets or one works. One cannot do both at the same time.” — Peter Drucker, The Effective Executive (1967)
I’ve been thinking about that line a lot over the last couple of weeks.
As some of you know, I recently stepped out of my role as CPTO at RightShip. One of the first things I did once I had a bit of breathing space was to look back and ask myself a simple question: “where did my time actually go?” Not where I thought it went, or where I wanted it to go, but where it really went.
The honest answer is that a lot of my last year or so was spent in meetings. And when I compare that with other roles I’ve had, especially as a founder and builder, I find myself wondering whether that was always the best use of my time.
So this article is me thinking out loud about the all-day meeting trap, why I think so many of us fall into it, and how thinking AI-first can help us stay out of it (or at least be a lot more honest with ourselves about it).
Is it just me? Nope!
Before being too hard on myself, I wanted to check whether this was just me or a much more common problem. It turns out it’s very common, and it has been getting worse for decades.
CEOs spend 72% of their work time in meetings — Porter and Nohria tracked 27 CEOs for 13 weeks: about 37 meetings a week, 70% of them an hour or longer, mostly out of habit.
Thinking time is scarce and chopped up — only 28% of their time was spent alone, and 59% of that came in chunks of an hour or less.
It’s been creeping up for years — executives now spend nearly 23 hours a week in meetings, up from under 10 in the 1960s.
Meetings make more meetings — at one company, a weekly executive meeting worth 7,000 hours a year grew to 300,000 hours once you counted the prep meetings beneath it.
Nor are people happy about it: Atlassian found 72% of meetings are rated ineffective, and 67% of directors and above work overtime because of them.
So… are you falling into the trap too? If you’re reading this between two back-to-back calls, you probably know the answer 😅
So what’s the magic number?
So what can we actually do about spending too much time in meetings, especially when you’re in a leadership role?
In product and tech, one thing we always advocate for is that you need to be able to measure. And before you can measure anything, you need to come up with some metrics that let you objectively check whether you’re really sticking to the outcome you wanted. Looking back, I never did that for my own time. We talk a great deal about how we measure our products, yet as managers we rarely measure ourselves with the same rigour, and that’s how habits form that probably shouldn’t.
Here are a few metrics I think are a good place to start:
% of your time in meetings — there’s no magic number, and Andy Grove rightly argued that meetings are how managerial work gets done. But Drucker’s rule of thumb was that a quarter of your time or more is a sign something is off, and Grove drew the same line for ad hoc meetings. So maybe 25% is a good start, but you’ll need to find your own number. Whatever that number is, what’s important is that you have one — a number you can continuously build towards, and that works for you and your business.
Outcomes from your meetings — did it actually move something forward? For a status or planning meeting that means a decision or an artifact (a plan, a PRD, a roadmap change). For an intro call with a partner or an interview with a candidate, it means a next step, an intro, a role, or even a clear “no” — all good outcomes. A meeting with none of these is usually a status update in disguise, and 77% of people say meetings mostly lead to more meetings. Hours in versus outcomes out gives you a hypothetical ROI on your time, and over time it shows you which kinds of meetings actually turn into opportunities.
How many of your meetings run long — 70% of CEO meetings ran an hour or longer, mostly out of habit, and Steven Rogelberg’s research shows a meeting expands to fill whatever time you give it. Make shorter the default (Outlook and Google Calendar both have a setting to shorten meetings automatically), and cap how many hour-long meetings you allow yourself each week.
If you take nothing else from this article, take this: there’s no right number, only the one that works for you — so pick it, and measure yourself against it.
Where I always fell short!
Now, I’m pretty sure all of this sounds great on paper and is really, really hard to keep up in practice. I know because I’ve done it myself many times. I’d get a burst of inspiration, set up the KPIs, start measuring the return, and for a week or two it was brilliant. Where I always fell short was consistently continuing to do it! Realistically, it just takes work, discipline and time you don’t have.
And sometimes that’s exactly the problem with executive roles: the people who most need this data have the least time to collect it. With a calendar already 70% meetings, categorising each one on a Friday evening just isn’t going to happen.
Even companies that really try find it hard if they don’t turn it into a system. Example: back in 2023 Shopify cancelled around 12,000 recurring meetings in one go. At first, meeting time dropped by a third, but within a few months it was only down 14%, still good, but half the gain had gone.
Meetings creep back. A one-off clean-up isn’t a system — you need something that keeps watching, week after week, long after the inspiration has worn off.
So how do we fix that?
Thinking about it AI-first
This is where I want to practise what I’ve been preaching for quite some time now — how would you approach this problem if you thought about it AI-first?
The discipline problem is exactly the kind of thing the agent revolution is good at. Not a chat window you need to remember to open now and then, but an autonomous, continuous, always-on agent that keeps an eye on a specific problem for you. You can build one today with OpenAI’s new dots (always-on agents with their own cloud computer, launched at DevDay in September) or with Grok Bot, which I wrote about recently and use every day.
And this is the bigger point I want to make. When you look at your work with an AI-first mindset, some of the best use cases are the things that used to depend on discipline, effort and time you simply didn’t have. Anything that needs consistency, and data you’ll never get round to collecting yourself, is an agent waiting to be built.
So that’s exactly what I did. I built an agent to systemise this for myself, so that whatever my next role looks like, I don’t fall back into the all-day meeting trap. I gave it a name and a bit of an identity, because it makes it feel like part of the “Marlon AI team” (yep I have a full team :) ) rather than just another tool.
Meet Tempo — an agent whose whole job is to set the pace of my week. Calm, direct and on my side: protective of my focus time, honest when I’m drifting, and gives me back time.
I published Tempo on Grok Bot — here is the link if you want to try it out.
https://x.ai/bot/OEHcyNENWnqn9VHAgW-tI
I designed Tempo around a simple loop it runs for me every single day, starting before my meetings rather than after them: Screen → Protect → Prepare → Measure → Review.
Screen — Tempo checks every invite as it lands, not in an end-of-day post-mortem. Is it worth my time? Could it be 25 minutes? Does it break my focus block? It flags it, and I decide.
Protect — my focus time is protected from day one, not only once I’m over target. Tempo holds a 2-hour block every morning and defends it, checking with me before declining or moving anything.
Prepare — two hours before any important call, it asks me in one line what I want out of it. Going in with a goal is how a meeting ends with an outcome.
Measure — it quietly logs every real meeting across my work and personal calendars (skipping the dentist or a dinner booking, so the % means something): how long it ran, whether it was ad hoc, and what it moved forward. Anything it can’t work out goes into my next morning digest.
Review — a short morning digest and a Friday review keep me on the hook: where I’m heading against my targets, what to shorten, decline or turn into a written update, and which kinds of calls actually turn into opportunities. That last one is where the real time savings are. I say yes in a quick chat, and Tempo does the rest.
I’ve only just started using Tempo, so I’ll report back properly. But even from the limited testing I’ve done, it has already saved me hours - suggesting meetings to shorten and drafting the email to the organiser for me, and spotting meetings with the same people and similar topics that could be merged into one. And that’s with my current calendar, which isn’t nearly as overloaded as it used to be!
All the hard work of managing my calendar stops being a burden on me. It becomes the sole focus of Tempo, an agent that’s purely dedicated to making my meeting time more efficient.
Does it solve the problem on its own? No — I still have to act on what it tells me. But it definitely makes me a lot more objective, and it does the boring, consistent part that I (and I suspect most of you) never managed to keep up.
And the cool thing about systemising this into an AI is that, over time, I keep giving it feedback, and because these agents have persistent memory they keep improving as time goes by. So really you end up with something highly tailored to your needs, yet fully automated and autonomous, helping you day by day at practically zero cost to your time. That, for me, is how AI should be applied in our everyday lives.
Build your own Tempo
Here’s the starting prompt I used for Tempo, so you can build your own in dots, Grok Bot or whichever always-on agent you use. Edit the targets to match your own numbers:
ROLE
You are Tempo, my meeting coach. Your one job is to set the
pace of my week: make the time I spend in meetings count, so
I have more time for the work that matters.
Personality: calm, direct and on my side. Protective of my
focus time, honest when I’m drifting, never pushy. Sign
every message “- Tempo”.
MY TARGETS (edit these)
- No more than 25% of my working week in meetings
- No more than 3 meetings of 60+ minutes a week
- A 2-hour focus block every working morning
- Every meeting moves something forward
WHAT COUNTS
- Look at both my work and personal calendars.
- Count only real meetings with other people. Leave out
personal items (appointments, dinners, the school run) and
my own focus holds.
WHAT “MOVED FORWARD” MEANS
- Status or planning meeting: a decision or an artifact
(plan, doc, PRD).
- Intro, partner, sales or candidate call: a next step, an
intro, a role, or a clear “no”.
- None of the above = no outcome.
ACCESS AND APPROVALS
- You can read my calendars, email and documents.
- Never accept, decline, move or message anyone without my
OK.
- Ask for approvals in this chat, not by email.
1. SCREEN (as each invite lands)
Check: is it worth my time against my targets, could it be
25 minutes, does it clash with my focus block? Flag anything
doubtful with a suggested reply. I decide.
2. PROTECT (from day one)
Propose a standing 2-hour morning focus hold. Once I
approve, defend it: flag any invite that lands on it and
suggest another slot.
3. PREPARE (2 hours before any important call)
Send me one line: “What do you want out of this?” Keep my
answer so you can check the outcome against it afterwards.
4. MEASURE (quietly, every evening)
For each meeting, log: length booked, attendees, recurring
or ad hoc, type of meeting, and what it moved forward (check
my email and docs). Don’t message me - save anything you
can’t work out for tomorrow’s digest.
Keep a running scorecard: % of my week in meetings, ad hoc
share, 60+ minute meetings, longest focus block, and
outcomes per meeting hour.
5. REVIEW
- Every weekday at 07:30, one short digest: where I am vs
target, today’s meetings worth questioning, and any open
questions.
- Every Friday at 16:00, a weekly review: scorecard and
4-week trend, recurring meetings with no outcome, and
which types of calls turn into opportunities and which
never do.
Lead with the numbers, be direct, and suggest specific
changes (shorten, decline, delegate, make it a written
update). When I say yes, do it and send a short, polite note
to the organiser.
LEARN
When I approve, reject or correct a suggestion, note why and
adjust. Keep a short list of my rules and preferences, and
show it to me in every Friday review.Treat this as your first version, not the final one. What really matters is that you keep refining it.
From the gym AI agents to the crazy calendar
In my last post I talked about how I use these bots for my gym sessions and my eating habits. What I find super interesting is that the exact same pattern works for a very business-centric problem like this one. The big takeaway for me is this: once you systemise something, it keeps getting better over time. What today seems unrealistic becomes a reality with the help of AI. Meetings are just one example - look for the habits you keep promising yourself you’ll stick to, and ask whether an agent could stick to them for you.
Drucker didn’t have agents in 1967, and I didn’t use one in my last role. Looking back, I wish I had — which is exactly why I’ve built Tempo for whatever comes next.
Special thanks to Claude and GrokBot for helping me write this article :)
First published on Medium on 7 October 2026.