Euler CenterWork Composition Run the assessment
Run it

How to run it in a room.

One workflow, one canvas, ninety minutes. Not the whole company. There is one way to run the assessment and sixteen reasons to run it, and this page keeps them apart: the protocol first, the applications second.

One protocol90 minutes, five moves
Sixteen applicationsFour groups: diagnose, decide, align, prove
One workflowNever the whole company
RuleA blank cell is a finding, not an embarrassment
Part I / How you run it

The protocol: five moves, ninety minutes

This is the one way to run the assessment. The bar shows how the time splits; the second move is the long block, and the one teams try to skip.

110 minutes

Pick one workflow, not the company

Name a single flow of work that already has an agent in it or is about to. Loan adjudication, tier-one support, supplier onboarding, evidence collection. Then ask the first question before any other: what result is needed, who needs it, and what work could we remove while still achieving it? A canvas filled at the level of the enterprise is a poster. A canvas filled at the level of a workflow is a decision.

235 minutesThe long block

Fill Proof first, not Intent

Everyone wants to start with purpose. Start at the Evidence cell instead and ask what you would hand an auditor tomorrow. It is the fastest way to find out how much of the operating model is currently a belief. Then work backward through Motion, Composition, and Intent.

315 minutes

Score honestly, three states only

Blank means nobody owns it. Thin means it was decided in a meeting and never written down. Solid means it is written, staffed, and produces evidence without being asked. Intention counts as blank. This rule is the whole method.

420 minutes

Argue about the disagreements, not the scores

Where legal scores a cell blank and engineering scores it solid, you have found the real gap, and it is a gap in shared understanding rather than in the system. Record the disagreement inside the cell. Do not average it away.

510 minutes

Leave with owners, not actions

Every blank cell leaves the room with a named person and a date, and nothing else. Twelve cells generate more actions than any team can carry, so cap it: the three blanks with the shortest path to an incident, and stop.

That is the whole protocol. Everything below runs these same five moves. What changes is why.See what you run it for
Part II / What you run it for

Sixteen applications, four groups

Which one you are running depends on who is holding the canvas. Each application below names 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.

Diagnose

Applications 01 to 04

Find out where the work actually stands before anyone proposes a change.

01

Baseline audit

Score all twelve cells for one live workflow. The output is not a maturity score, it is a list of blank cells with names against them.

In the room
Workflow owner, engineering lead, risk or compliance, one person who does the work
Time
90 minutes
Output
A blank-cell list with owners
Fails when
You run it at company level and produce a poster nobody acts on

Cells: All twelve

02

Blank-cell postmortem

After an incident, do not ask what broke. Ask which cell was blank when it broke. Every case in the library resolves to one.

In the room
Incident responders and the workflow owner. No executives in the first pass
Time
60 minutes
Output
A cell attribution and one systemic fix
Fails when
It becomes a search for the operator who missed it, which is the exact failure the Evidence cell exists to prevent

Cells: Usually Authority, Evidence, or Coordination

03

Cultural self-location

Plot where the organization actually sits on the traditions map, then where the new operating model assumes it sits. The distance is your change cost.

In the room
Leadership team, ideally one person per region you operate in
Time
45 minutes
Output
Two points and an honest conversation about the gap
Fails when
It turns into a values exercise instead of a design constraint

Cells: Purpose, Authority, Incentives

04

Delayering pre-mortem

Before removing a management layer, list what that layer actually does under Coordination, Mastery, and Structure. Whatever you cannot reassign becomes homeless work.

In the room
The executive proposing the cut, plus two managers from the layer being cut
Time
2 hours
Output
A reassignment map, or a smaller cut
Fails when
It is run after the decision, as justification

Cells: Structure, Coordination, Mastery

Decide

Applications 05 to 08

Settle a live question with the scored canvas in the room.

05

Work composition triage

Run each task through the placement matrix. The single highest-value hour in the method, because it is where the Klarna mistake gets made or avoided.

In the room
Two people who know the work in detail, one who knows the models
Time
2 to 3 hours per workflow
Output
A task-level automate, hybrid, or human decision with the reasoning recorded
Fails when
Everything gets classified complicated, because complex feels like an admission of ignorance

Cells: Composition, Authority

06

Build, buy, or orchestrate

Place each capability on the evolution curve and let position dictate sourcing, rather than letting a vendor relationship dictate it.

In the room
Architecture, procurement, the business owner
Time
Half a day
Output
A sourcing decision per capability, with a review date
Fails when
Genesis capabilities get procured and commodities get lovingly built in house

Cells: Strategy, Resources

07

Investment allocation

Fund the blank cells, not the demo. Most agent budgets go to Composition while the failures happen in Authority and Evidence.

In the room
Whoever controls the budget, plus risk
Time
One planning cycle
Output
A funded gap list ranked by shortest path to an incident
Fails when
Cells with no visible demo attached lose to cells that have one

Cells: All twelve, weighted by blanks

Read the funding model
08

Portfolio comparison

One canvas per business unit or workflow, side by side. Patterns appear fast: the same cell blank everywhere is a platform problem, not a local one.

In the room
Operating model owners from each unit, together, once
Time
Half a day
Output
The list of cells blank in most units
Fails when
It becomes a league table and units start scoring themselves generously

Cells: All twelve, across units

Align

Applications 09 to 12

Get two parties describing the same work in the same words.

09

Shared language across functions

The most valuable thing the canvas does. Legal, engineering, HR, and compliance stop arguing past each other because the argument now has twelve named boxes.

In the room
All of them, which is the point
Time
Recurring
Output
Fewer meetings about what the meeting is about
Fails when
One function owns the canvas and it becomes their document

Cells: All twelve

10

Executive and board onboarding

A new CTO, CHRO, or chief risk officer fills the canvas in their first fortnight. Their blanks are more honest than the incumbents', because they have nothing invested yet.

In the room
The new executive plus two people who will tell them the truth
Time
One session, then a follow-up
Output
An outsider's blank-cell list, which is the useful one
Fails when
They are handed a completed canvas to read instead of an empty one to fill

Cells: All twelve

11

Vendor and partner diligence

Make the vendor fill your canvas for their product. The cells they cannot fill are the ones you will fill yourself, at your own cost.

In the room
Procurement, architecture, and the risk owner who inherits it
Time
90 minutes with the vendor
Output
A vendor canvas with your inherited gaps marked
Fails when
You accept the Composition and Flow answers and never press on Evidence and Authority

Cells: Composition, Evidence, Authority, Resources

12

Merger and integration fit

Sketch both operating models and look for incompatible cells rather than overlapping functions. Two firms can have identical org charts and irreconcilable Authority cells.

In the room
Integration lead and one operating model owner from each side
Time
Two sessions
Output
A ranked list of incompatible cells
Fails when
Diligence covers the balance sheet and the headcount and never the decision rights

Cells: Authority, Incentives, Coordination, Purpose

Prove

Applications 13 to 16

Produce the record an auditor, a board, or a regulator will ask for.

13

The go-live gate

No agent touches regulated work until six cells are solid. The version that gets bought, because it converts a framework into a control a second line of defense can enforce.

In the room
Risk, the deploying team, and whoever signs
Time
30 minutes per deployment
Output
A pass, a conditional pass with a date, or a stop
Fails when
It becomes a form. Gates that are never failed are not gates

Cells: Authority, Composition, Flow, Evidence, Mastery, Incentives

14

Audit and framework mapping

Map the Evidence cell once to the regimes you live under, then reuse it. The overlapping frameworks ask similar questions in different vocabularies.

In the room
Compliance, plus the engineer who has to produce the artifacts
Time
Two to three days once, then maintenance
Output
One control set, several framework mappings
Fails when
You map to the frameworks and never test whether the system can actually produce the evidence unprompted

Cells: Evidence, Authority, Mastery

15

Role design and credentialing

Each cell already names a capability. Turn the twelve into assessed roles and you have the credential nobody currently issues.

In the room
Capability lead, assessors, a regulator-facing reviewer
Time
A program, not a session
Output
Assessed roles rather than attended training
Fails when
It becomes a course with a certificate instead of a simulation with a pass mark

Cells: Mastery, Authority, Evidence

16

Quarterly instrument panel

One indicator and one threshold per cell, reviewed on a cadence. Cycle time for Flow, override rates for Authority, evidence completeness for Evidence.

In the room
Operating model owner, quarterly
Time
One hour a quarter
Output
Twelve trends and one cell to work on next
Fails when
Indicators get chosen because they are easy to collect rather than because they would change a decision

Cells: All twelve