Finally, Meetings That Move Your Company Forward

Most meetings are theatre. A lot of talking, and at the end nobody is any wiser.

I work at the intersection of product, GTM and AI, and I build with small teams. Three people instead of thirty. In a team like that you cannot afford a meeting where two hours pass without a single usable suggestion. The problem is almost never the team. It is the missing structure.

What you take away:

• Why a three-person team decides instead of meets.
• How I structure meetings with the IED principle.
• How AI agents take over the prep so the humans only decide.

My thesis: a meeting exists to decide, not to inform

A good meeting needs three things: a rhythm, a driving engine and process discipline. The rhythm is the regularity, like a heartbeat. The engine is the commitment, yours included. Process discipline keeps you on the agenda. Drop any one of them and you go in circles.

🧨 The problem: called on a whim, runs without structure

The classic pattern: you want to make a decision, you call a meeting spontaneously, everyone talks at once, new topics keep popping up, and there are people at the table who cannot contribute. An EOS survey found that executives rate the effectiveness of their meetings at an average of 4 on a scale of 1 to 10. That is the reality.

The costs of this never show up on an invoice. An unstructured meeting with 5 people does not eat one hour, it eats 5. Add the ramp-up afterwards until everyone is back into their work. And the creeping effect is worse than the direct one: whoever sat through 3 meetings that decided nothing stops preparing for the fourth. That is how structure dies before it exists.

🛠️ How I build meetings

First I classify every topic with the IED principle:

• I = Information. Share only, no discussion.
• E = Decision (Entscheidung). This is where we decide.
• D = Debate. This is where we wrestle.

That way everyone knows in advance what happens at each point. As a skeleton I use the Level 10 agenda from the EOS framework. Compressed to 45 minutes, a run looks like this for me:

• Check-in, one to two minutes. What went well.
• Updates from the areas, five minutes.
• Customer and team headlines, one sentence each, five minutes.
• To-do list, five minutes. Done, next week.
• Problem list, twenty minutes. The three most important items, focus on solving.

Plus the ground rules: start on time, hard stop, only relevant people. Too many cooks spoil the broth. In a small group everyone gets to speak, and that makes the meeting better instantly.

Meetings by the rules of science

Jacco van der Kooij describes a science culture in his SaaS Sales Method, as the counter-model to superstar culture. He means sales. But his principles are the best meeting rules I know. I use 3 of them directly.

First: decisions from data. Measurability creates accountability, says van der Kooij. In the meeting that means: if you have an opinion, show a number. Not because opinions are worthless. But because a number shifts the discussion from the loudest voice to the provable. It does not work without a clean data foundation, more on that in my piece on data strategy in the CRM.

Second: team performance over individual performance. The meeting looks at the team's numbers, not at a ranking of heroes. When a metric is red, nobody asks who is to blame. We ask which step in the process is stuck. That takes the theatre out of the room. People stop defending themselves and start thinking along.

Third: an iterative cadence. Van der Kooij recommends changing one variable and reviewing the result in intervals of 15 to 30 days. One variable, not 5. A meeting that decides 3 measures at once learns nothing, because nobody knows which one worked. So: one change per cycle, one review date, one clear result. That is slower per week and faster per quarter.

Together the 3 rules produce a meeting that behaves like an experiment: hypothesis, measurement, decision. Which fits the IED principle from above exactly. Information is the measurement, debate is the interpretation, decision is the next experiment.

Which meetings do you actually need?

3 are enough. A weekly numbers review on a scorecard, a fixed list of 5 to 7 metrics. A decision meeting with prepared briefs, meaning one-page decision documents. A monthly retro. Everything else runs asynchronously. A small B2B software company does not need more.

The numbers review, weekly, 30 to 45 minutes. One scorecard, 5 to 7 metrics, a name next to each. Green or red, no storytelling. For example pipeline, replies to outbound (your active outreach), active users, churn signals (early warnings that customers are leaving), cash. Whatever is red moves to the problem list. It does not get solved in the review, only spotted. That separation disciplines more than any facilitation technique. The review can run as the compressed Level 10 agenda from above, the core stays the scorecard.

The decision meeting, as needed, but never without a brief. Every topic arrives with one page: context, options, recommendation, cost, benefit. No brief, no slot. That sounds bureaucratic and is the opposite. Whoever writes the page thinks first. And the group decides in 10 minutes what it would otherwise have discussed for 60.

The retro, monthly. 3 questions: what did we change, what did it do, what do we change next. That is van der Kooij's cadence, translated into the calendar. Without a retro every experiment fizzles out, because nobody collects the result.

And the rest? Asynchronous. Status updates, project states and FYI topics belong in the tool, not in a room. The IED test from above applies hard here: what needs neither a decision nor a debate needs no meeting. To me that is part of the craft of building a company: calendars are cost centres.

🤖 The tool: agents do the prep

Here is the 2026 part. In a Hyperlean team, a few pros plus AI agents that work around the clock, the prep is no longer human work. The agents pull the updates from the system, summarise the customer headlines, prepare the three problems with context and propose options. All of it runs in GTM OS, the app on app.gtm.science you use to manage tasks and context, like a Jira for GTM teams.

For that to work, the agents need a foundation: the Context Engine, your company's business context as the basis for AI agents. Without that context it is garbage in, garbage out. With it, the information column of your meeting is ready before the meeting. The humans walk in and do the one thing humans do better: decide.

Concretely, the prep looks like this. Before every meeting the agent assembles 3 things. First the numbers: the scorecard with current values and the deltas to last week, marked where it hurts. Second the open decisions: each item as a short brief with context, options and a recommendation. Third last week's commitments: who promised what, what is done, what is open.

The result is one page everyone reads before the meeting. In the meeting itself nobody reports any more. There is nothing to report, it is all on the page. The 45 minutes belong to what agents cannot do: weigh up, take responsibility, decide. After the meeting the agent writes the decisions and the new commitments back into the system. Next week's prep starts itself.

The effect is unspectacular and big at the same time. Not because the agents are smarter than your team. But because they take over the collecting that used to eat the first half hour of every meeting.

The strongest argument: less meeting, more output

The biggest mistake is holding no meetings at all. The second biggest is holding them only on demand, until the need for clarification piles up. Regularity beats both. But regularity plus agents that deliver the information beats everything. When the prep is done, the meeting shrinks to the part that really needs humans. Three people, a clear agenda, a decision. Instead of thirty people listening.

There is one more effect I underestimated for a long time: decisions get a history. When you decide on the same scorecard every week, after a quarter you can see which decision did what. That is the difference between a team that remembers and a team that reruns every discussion from scratch.

🎢 Highs, lows, warnings

✅ What works: a fixed rhythm, IED classification, agents for the prep. The meeting turns from a time sink into a decision tool.

❌ What does not work: wasting the meeting on the new coffee beans or the empty toilet paper roll. And grabbing the entire facilitation for yourself.

⚠️ Warning: agents without context deliver plausible fog. Garbage in, garbage out. The prep is only as good as the context it runs on.

Back to the start: meetings do not have to be theatre. With rhythm, IED and agents that bring the information, the meeting finally becomes what it should be. The place where your company decides and moves forward. Three instead of thirty.