Pattern AI #13: Anatomy of a Pattern

Pattern AI #13: Anatomy of a Pattern

In the previous post, we explored the idea of a Pattern Curriculum for Accomplishment in the AI Age.

But before patterns can become units of learning, application, and deployment, we must answer a foundational question:

What exactly constitutes a useful pattern?

A recurring observation alone is not yet a complete pattern.

For a pattern to become a leverage point, it must tell us where it applies, what activates it, how it works, what action it enables, and how its value can be judged.

A Pattern Is Compressed Intelligence

A pattern captures something that recurs across situations.

It may represent:

• A relationship between variables

• A recurring problem and response

• A structure within information

• A sequence of actions

• A decision rule

• A method for producing an outcome

• A warning that indicates risk

• A reusable form of human–AI collaboration

An AI-generated pattern is like a mechanical spring: compressed knowledge, experience, relationships, and possible actions, ready to release value when activated in the right context.

However, compression alone is insufficient.

A useful pattern must also be understandable, actionable, testable, adaptable, and governable.

The Anatomy of a Pattern

1. Name

A pattern needs a short and memorable name.

The name enables people to recognize, retrieve, discuss, and reuse it.

For example:

• Compare Before You Choose

• Evidence Before Conclusion

• Decompose Before Delegating

• Human Review at High-Risk Points

A good name turns an invisible structure into a shared concept.

2. Purpose

What problem does the pattern address?

The purpose explains why the pattern exists and what it is intended to make possible.

Without a clear purpose, a pattern can become a technique in search of a problem.

3. Context

Where does the pattern apply?

Context may include the domain, task, learner, organisation, constraints, available information, and level of risk.

A pattern that works well in one context may be ineffective or harmful in another.

4. Signal or Trigger

What indicates that the pattern is needed?

The trigger could be:

• A recurring problem

• A decision involving multiple options

• Missing or uncertain information

• An inconsistency in the evidence

• A repetitive task

• A high-risk action requiring human review

A pattern becomes useful when people can recognize the moment to activate it.

5. Inputs

What does the pattern need before it can work?

Inputs may include:

• A defined outcome

• Relevant data

• User requirements

• Constraints

• Examples

• Evaluation criteria

• Domain knowledge

Poor or incomplete inputs can weaken even a well-designed pattern.

6. Structure

What is the recurring relationship, logic, or sequence within the pattern?

This is the pattern’s internal mechanism.

It may take the form of:

• If–then logic

• A sequence of steps

• A comparison framework

• A feedback loop

• A classification structure

• A problem–response relationship

• A set of interacting roles

The structure explains how the pattern converts inputs into action.

7. AI Role

What should AI do?

AI may:

• Discover relationships

• Recognize known structures

• Retrieve and organise information

• Generate alternatives

• Compare options

• Simulate possibilities

• Detect inconsistencies

• Produce a draft artifact

The AI role should be explicit rather than assumed.

8. Human Role

What must remain under human judgement or responsibility?

The human may need to:

• Define the purpose

• Provide context

• Select the appropriate pattern

• Examine evidence

• Resolve ambiguity

• Consider values and consequences

• Approve high-impact decisions

• Accept responsibility for deployment

A Pattern AI unit should clarify both AI capability and human accountability.

9. Action

What should happen when the pattern is activated?

The action converts the pattern from knowledge into work.

It may instruct the human, the AI, or both to perform a defined sequence of activities.

10. Artifact

What tangible output should the pattern produce?

The artifact may be:

• A decision canvas

• A diagnosis

• A plan

• A design

• A comparison matrix

• A research brief

• A reusable prompt

• A workflow

• An agent

• An impact report

The artifact makes the application of the pattern visible and reviewable.

11. Expected Outcome

What should become possible after applying the pattern?

The outcome must express the intended improvement, such as:

• A better-informed decision

• Faster completion

• Reduced error

• Improved consistency

• Earlier risk detection

• Greater personalisation

• A reusable organisational capability

The artifact is what was produced. The outcome is the value it created.

12. Evidence

How will we know that the pattern worked?

Evidence may include:

• Accuracy

• Time saved

• Quality improvement

• User acceptance

• Reduction in errors

• Achievement of defined criteria

• Comparison with a baseline

• Successful reuse in another context

Without evidence, a pattern remains a plausible idea rather than a validated capability.

13. Boundaries

When should the pattern not be used or trusted?

Boundaries may arise from:

• Insufficient evidence

• Poor-quality data

• Changed conditions

• High-stakes consequences

• Legal or ethical restrictions

• Contexts outside the pattern’s original scope

• Situations requiring qualified professional judgement

A responsible pattern explains not only where it works, but also where it may fail.

14. Transfer

Where else can the pattern be applied?

A valuable pattern can often move:

• From one task to another

• From one subject to another

• From an individual to a team

• From a prompt to a workflow

• From a workflow to an agent

• From a local practice to organisational capability

Transfer is one of the principal sources of a pattern’s leverage.

15. Learning and Evolution

How will the pattern improve through use?

Patterns should not be treated as permanently finished.

Application produces evidence. Evidence reveals strengths, limitations, and new possibilities. These insights can then improve the pattern.

The lifecycle becomes:

Use → Observe → Judge → Refine → Reuse

A Simple Example

Consider the pattern Compare Before You Choose.

• Purpose: Improve decisions involving multiple alternatives.

• Context: A choice must be made among competing options.

• Signal: Several options appear attractive, but their trade-offs are unclear.

• Inputs: Options, criteria, constraints, priorities, and available evidence.

• Structure: Define criteria → assign importance → compare options → identify trade-offs.

• AI role: Organise information, generate comparisons, and expose overlooked differences.

• Human role: Set priorities, validate evidence, consider consequences, and make the final decision.

• Action: Evaluate every option against the same criteria.

• Artifact: Decision comparison matrix.

• Outcome: A more transparent and defensible decision.

• Evidence: The selected option satisfies the most important criteria and withstands review.

• Boundary: The pattern should not replace professional judgement in high-risk decisions.

• Transfer: It can be used for selecting a course, supplier, technology, investment, candidate, or strategy.

This example shows how an ordinary idea becomes a complete and reusable Pattern AI unit.

From Pattern to Accomplishment

A complete pattern creates a bridge between recognizing a recurring structure and producing a meaningful result.

Name makes the pattern retrievable.
Context and signal make it activatable.
Structure makes it understandable.
Roles and actions make it usable.
An artifact makes it visible.
Evidence makes it credible.
Boundaries make it responsible.
Transfer makes it a leverage point.
Learning keeps it alive.

This leads to a working definition:

A pattern is a named, contextual, reusable and testable structure that connects a recurring situation to coordinated human–AI action and a measurable outcome.

Understanding the anatomy of a pattern is essential for building a Pattern Curriculum.

Only when patterns are clearly structured can learners recognize them, apply them, adapt them, combine them, create new ones, deploy them, and govern their use.

Pattern is the unit of learning.
TAJ is the method of engagement.
Artifact is the evidence of application.
Outcome is the proof of accomplishment.
Reflection improves the pattern.

#PatternAI #PatternCurriculum #AIYING #TAJ #AIInEducation #FutureOfLearning #Accomplishment