▶️ Running Plays

How to Run a Play

  1. Define the Problem Context Start with the user workflow or system pain point you're solving.

  2. Select Interaction Modes and Patterns Identify the relevant interaction paradigms, UI patterns, and input/output modalities from Sections I.

  3. Choose the Integration Approach Decide whether this Play fits as Embedded AI, Standalone AI, or Cross-System Integration.

  4. Map the Archetypes Specify what agents, tools, and workflows will be used.

  5. Set key Dimensions Explicitly define key system attributes such as autonomy level, context awareness, workflow role, trigger source, etc.

  6. Decide Runtime Behaviors Choose how AI will act at runtime: passive, reactive, or proactive.

  7. Assign Human Roles Clarify who builds, configures, and operates the system.

  8. Layer in Prompting & Evaluation Strategy Specify how system prompts will guide AI behavior and which evaluation methods will measure quality.

  9. Apply Design Considerations Ensure the Play builds trust, gives control, and fits into user mental models.

  10. Design Outcome Create example UI flow, system diagram, or user journey.

Play Template

Play Name:

Problem: What task or challenge does this solve?

Interaction Model: What’s the interaction model?

Integration Model: Embedded, Standalone, Cross-System?

Archetypes: Which agents, tools, workflows are involved?

Dimensions: Autonomy level, context awareness, trigger source, etc?

Runtime Behaviors: Passive, Reactive, or Proactive?

Human Roles: Who configures, operates, and interacts with it?

System Prompts: How are prompts designed to guide AI behavior?

Evaluations: What evaluation methods ensure quality and drive iteration?

Design Considerations: Key UX or system design concerns.

Example Outcome: Example diagram or flow.

Updated on