The Work Composition Canvas teaches you how to systematically understand, design, and govern how people and agents share your work.
Twelve decisions every operating model is already making, most of them by accident. One surface to make them visible, one gate to stop the bad ones shipping, and an assessment that says who is qualified to overrule the machine.
Published by The Euler Center under Creative Commons. Free to use, teach, translate, and adapt with attribution.
Nothing here assumes you have read the books. Every name is a link to an entry that explains it in plain language and says where the claim came from. If a term is underlined, it has a page.
The canvas belongs to everybody, including the consultancies that will compete with us using it. That is the point. A method nobody else will teach is not a standard, it is a sales deck.
Score twelve cells against one real workflow in ninety minutes. Blank, thin, or solid. The output is a list of blank cells with names against them, not a maturity grade.
Deconstruct the work, place each task by problem type, set the loop boundaries, and design the escalation that fires by itself. Six techniques, each with its failure mode.
Six cells solid before an agent ships, the Evidence cell mapped to the regimes you live under, and an assessment that names who may overrule the machine.
A blank cell is a finding, not an embarrassment.
Intent sets what the system is for and who may act. Composition decides what work exists and who holds it. Motion is how work moves and improves. Proof is what survives contact with an auditor. Fill it for one workflow, never for a whole company. Tap a cell, then any underlined name inside it.
Three points per cell. Blank means nobody owns it. Thin means it was decided in a meeting and never written down. Solid means it is written, staffed, and producing evidence without being asked. One anchoring rule holds everywhere: intention counts as blank.
A score with no record behind it is an opinion with a number attached. Five fields, kept per cell, make a score comparable across units and defensible a year later when nobody remembers the conversation.
Long Tail and Multi-Sided Platform entered the language because they were named shapes people recognized their own company in. These are the operating model equivalents, each drawn in the twelve cells, each with the cells it loads and the cells it starves. This library is open, and it is where we most want your submissions.
No agent reaches production touching regulated work until six cells are solid. Deliberately shorter than the canvas, because a gate people can hold in their head is a gate that gets used. Run it per deployment, not per program. Free, no signup, forward it to whoever needs it.
Map the Evidence cell once and reuse it. These regimes were written separately and converge on the same handful of demands: a person who can intervene, a record of why the system did what it did, and a management system that survives an auditor rather than a demo.
Evidence is the cell organizations most often score solid and are most often wrong about. These four practices separate a system that produces evidence from a team that can reconstruct it under pressure.
The canvas is free and deliberately uncalibrated. Where the line falls between thin and solid depends on your regulatory regime, your workflow, and what it costs you to be wrong, so we do not set it centrally. That is not an omission with a paywall behind it. It is the reason the rungs below exist at all: calibration is how the method completes, and it is local work that somebody has to do.
Every rung is marked with what actually exists today.
The canvas only works as a shared language if the language is actually shared. Legal, engineering, and compliance arguing past each other is usually a vocabulary problem wearing a governance costume. These are the forty entries turned into cards. Free, no signup, and self-scored, because nobody is grading you.
Concepts are the ideas the canvas inherits, explained without assuming you have read the source. Cases are what actually happened at named organizations, with the reporting behind each claim and how strong that reporting is. Nothing here asks you to take a number on trust.
Draw an operating model shape you have actually run or seen, in the twelve cells.
Submit an operating model failure, anonymized, attributed to a blank cell.
Correct the cultural entries in public. This is the part most likely to be wrong from a desk in Texas.
Being specific about what is missing is more honest than a general call for collaborators, and easier to say yes to.
Twelve cells. Seven shapes. Forty entries. Come argue with any of them.
The canvas describes an operating model. These are the moves you make when you have to design one rather than document it. Each borrows from a discipline that already solved a version of the problem, and each names where that discipline gets it wrong.
The hatched cells are where the documented failures happened. Not because the technology was inadequate, but because somebody put an agent where the problem was still being discovered.
Three instruments. A century of people solving these same twelve problems with slower machines. Four series that say the management layer is being removed faster than the capability to run without it is being built. And eleven traditions, because every firm is already running one of them, usually without having chosen.
Drag sideways to move through the century. Colored by lineage: quality and flow, cybernetics and self-management, the agentic era, and the failure record.
Eleven management logics on the two axes that actually separate them: how a decision gets made, and whose interest the organization is held to serve. Plot where you are, then plot where your new operating model assumes you are. The distance between those two points is your change cost. Tap a point.
One workflow, one canvas, ninety minutes. Not the whole company. The discipline that makes this work is the same one that makes an incident review work: a blank cell is a finding, not an embarrassment, and nobody gets to fill one with intention.
Which one it is depends on who is holding it. Each application carries the room it needs, the time it takes, what comes out, and how it goes wrong. That last field is the one most method libraries leave out.