Mark Eddleston's book The Pigeon Test makes one argument, and it is hard to unsee. Put a ball in a catapult and you can predict where it lands. Put in a pigeon and you can't. Most of what gets sold as transformation is calculated as a ball and launched as a pigeon.
His five free workshops, built in the Liberating Structures style, let a group find that out from its own work. The first one, Same Plan, Ten Times, asks one question about every piece of work on a wall: if we did this ten times with the same plan, would we get the same result every time?
We ran into this book at the right moment. At Paracosm, we work on a close question from another side: what happens to a learning organization when every person works beside agents that act?
Agents are very good at the ball
Work that gives the same result twice is exactly where agents shine. Booking, routing, drafting from a template, reconciling, reporting. Put it on the left of the wall, give it to agents, supervise them, and stop spending human attention on it.
The risk is on the right. Work that would not give the same result twice, such as a restructure, a culture shift or a new service, cannot be handed to an agent with a plan and a date. An agent will execute a plan faster than anyone, which means it will also execute a bad plan faster than anyone. Speed does not turn a pigeon into a ball.
So the first use of Mr. Eddleston's wall in the agentic era is practical: sort your work before you automate it.
What the right side needs is learning, not a plan
On the right, the useful move is to probe, read what comes back and adjust. That is what Peter Senge described as a learning organization: people who keep expanding their capacity to create the results they want, through systems thinking, shared vision, mental models and team learning.
Paracosm's framework, Calm Magic, is built for that right side. It starts from a refusal that Mr. Eddleston would recognize:
- The future is a cone, not a line. Several futures are possible, fewer are plausible, fewer still are probable, and one is preferable. You do not predict one. You choose which to aim for.
- Habits beat plans. Three small habits do the work: name the friction (GL!TCH), loosen the framing (DRIFT), and turn what you see into a commitment others can act on (TUNE).
- Change is moves, not steps. Explore, narrow, commit. Skipping a move is the common mistake, and repeating one is allowed.
- The goal is that you no longer need us. The book is free and meant to be used without a consultant in the room.
Where we should be honest
There is a fair objection, and the book raises it about itself: a framework that comes with named stages can become the next recipe. Ours has stages, tiles and a vocabulary. Used as a checklist, it would fail the Pigeon Test.
Our answer is a rule, not a defence. The stages are prompts for a group to run on its own evidence, and the output is a commitment to test, not a roadmap. If a group uses it to promise a result by a date for work that is a pigeon, it is being misused. Mr. Eddleston's workshops, like ours, are early and get corrected as groups run them. We think that is the right posture.
The three together
- The Pigeon Test tells you which kind of work you have.
- A learning-organization practice tells you how to work with the pigeons: small probes, honest reading, shared commitments.
- Agents take the balls off your plate and make each probe cheaper and faster, under human supervision.
Crewdle, the platform behind this work, provides the agents. The part that makes an organization wiser, and not just faster, is still the group in front of the wall.
The theme: the learning organization
Underneath all of this is one idea. An organization that learns can work with pigeons. One that only plans cannot, with or without agents. That is why we put the learning organization at the center of everything we build.
Speed does not turn a pigeon into a ball.
Try it this week
1. Run Same Plan, Ten Times with eight or more people who work together. It takes an hour, a wall, sticky notes and dots, and it is free on Mr. Eddleston's site.
2. Take one dotted piece of work from the right. Name the friction in it, frame it three ways, and agree on one small test for next week.
3. Pick one piece of work from the left and ask whether an agent should own it.
Then tell us what the wall said.
Go further
- Discover the Calm Magic framework: https://paracosm.helloarchitekt.com/en/learning-organizations/
- Learn more about Mark Eddleston and The Pigeon Test: https://thepigeontest.com
- Learn more about Crewdle, the platform behind our agents: https://crewdle.ai
Paracosm is the learning-organization practice behind Calm Magic. Building Learning Organizations is our free book, in English and French.