Why Use This Playbook?
Designing for AI introduces conflicting assumptions, jargon, and mental models. This playbook gives teams a shared vocabulary and a step-by-step way to make clear, testable decisions fast.
When to Use This Playbook?
-
Exploring a new AI feature or product to frame options clearly.
-
Aligning a cross-functional team to establish shared vocabulary and decision flow.
-
Auditing an existing AI experience to check for gaps in trust, control, and UX.
What is a Play?
A Play is a structured recipe for designing an AI feature or system across the following building blocks.
-
🎛️ Interaction Models - How users engage with the AI.
-
🔌 Integration Models - Where AI fits in your product ecosystem.
-
🧱 Archetypes - Which components (agents, tools, workflows) are in play.
-
💠 Dimensions - How capabilities and properties (autonomy, context, scope) are defined.
-
⚙️ Runtime Behaviors - How AI behaves when operating.
-
👨💻 Human Roles - Who configures, monitors, and interacts with the system.
-
📃 System Prompts - How the system is instructed to behave.
-
🎯 Evaluations - How the system is measured and controlled.
-
📐 Design Considerations – Patterns for trust, transparency, and control.
How to Use This Playbook?
-
Compose a Play by selecting elements from each section.
-
Apply Design Considerations to ensure trustworthy, user-centric experiences.
-
Use the Play Template as your checklist