Skip to main content

Constraint Map

What in your business is actually stopping growth — and is that constraint real, or is it a process artifact?

Every business accumulates constraints the same way. Hit a capacity limit. Hire a person to solve it. Move to the next bottleneck. Repeat for ten years. The result: a workflow built from sequential hiring decisions, none of which was designed with the full picture in mind. Leaders internalize these constraints as fixed — "that's just the cost of doing business" — because nobody had an alternative. Now there is one.

The Constraint Map forces the most important question in AI transformation: if tripling the number of people working on this would unlock the bottleneck, it's an artifact. AI replaces artifacts. It cannot replace judgment.

0. Framing​

Before mapping, anchor to what the business delivers.

QuestionAnswer
What does this business deliver? (outcomes only — not how)[e.g. "acquires high-performing properties" not "runs an analysis team"]
If you 10× volume overnight, where does the first breakdown occur?[name the specific workflow, not the department]
What does "growth constraint" look like here — is it capacity, speed, quality, or cost?[pick one]
Who has the authority to eliminate a role or redesign a process?[must be named before Step 5]

1. Workflow Inventory​

List every workflow in scope. Do not filter yet. Include the ones everyone says are fine.

#Workflow / RoleWho Does ItVolume / WeekTime per UnitAnnual Hours
1
2
3
4
5
6
7
8
9
10

2. Classification​

Apply the tripling test to every workflow above. Column definitions:

TypeDefinitionTell-tale sign
ArtifactLabor-based. If tripling headcount would unlock the constraint, AI can replace it entirely.Junior roles exist to handle volume. Senior time spent on steps that require no judgment.
RealJudgment-based. A human must put their name on the outcome. Quality degrades without experience.Error in this step costs the relationship or the deal. Tacit pattern recognition is irreplaceable.
HybridContains both. The data-gathering or formatting layer is artifact. The read-between-lines layer is real.The step has a mechanical sub-task and a judgment sub-task — separate them.
#WorkflowTypeJudgment Layer (if Hybrid/Real)Artifact Layer (if Hybrid/Artifact)
1Artifact / Real / Hybrid
2Artifact / Real / Hybrid
3Artifact / Real / Hybrid
4Artifact / Real / Hybrid
5Artifact / Real / Hybrid
6Artifact / Real / Hybrid
7Artifact / Real / Hybrid
8Artifact / Real / Hybrid
9Artifact / Real / Hybrid
10Artifact / Real / Hybrid

3. Reality Test​

For each workflow marked Artifact or Hybrid, challenge the assumption that the constraint is real.

#WorkflowCurrent assumption ("that's just...")Is it actually fixed?Evidence or source of assumption
YES / NO / UNKNOWN
YES / NO / UNKNOWN
YES / NO / UNKNOWN

Common false constraints to surface:

  • "We need senior people to review all outputs" — often a trust deficit, not a quality requirement
  • "That's the nature of our industry" — usually an unexamined legacy process
  • "We tried automating it once and it failed" — failed attempts at bolt-on automation ≠ failed AI-native redesign
  • "Our clients expect a human to do this" — test this assumption directly; it is often not true

4. Constraint Priority​

Rank artifact workflows by transformation value. Score each on three dimensions (1–5).

WorkflowVolume (how often)Senior Time Trapped (how much leverage)Strategic Unlock (what it unblocks)TotalPriority

The binding constraint — the single workflow that, if eliminated, frees the most senior judgment capacity:

[Workflow name] — [one sentence on why it is binding]

This is the first module to attack. Everything else is sequenced behind it.

5. Before / After Picture​

For the top 3 workflows marked for transformation, sketch the before and after state.

WorkflowBefore (current)After (AI-native)What disappearsWhat remains human
1
2
3

The "what remains human" column is where judgment lives. This column defines the job of every senior person post-transformation — it must be worth doing.

6. Prerequisites Checklist​

Before any AI system can be built against an artifact workflow, verify:

  • Business logic is documented — the rules that govern this workflow are written down, not in someone's head
  • Success criteria are defined — there is a measurable standard for what "good" looks like
  • Context is available — the institutional knowledge, history, and policies needed to perform this task exist in a loadable form
  • Volume justifies build — workflow occurs at least ~25 times per month (below this, build cost rarely pays back)
  • Integration cost is known — API access, data connectors, and system touchpoints are costed
  • Change management is planned — the people whose artifact work will be replaced know what happens to them

Any unchecked box is a prerequisite, not an obstacle. The Business Logic Document and Context Architecture address the first two.

Context​

Questions​

When you classify a workflow as "Real" because it requires judgment — are you certain, or is that what you told yourself when you hired the last person to do it?

  • Which workflows in your business exist only because a previous workflow required a human hand-off?
  • What would your most senior person work on if every artifact task below them was handled?
  • Where does the constraint map reveal that the real bottleneck is not capacity — it is process design?