TPS 101: Foundational Framework
Moving beyond one-way commands to structured AI collaboration.
Welcome to Collaborative Prompting
Traditional prompting is often a one-way conversation. You ask a question, and the response is frequently too generic or misses the mark. This happens because most users talk at an AI, rather than with it.
What Makes TPS Different?
TPS is a collaboration framework. It provides a shared vocabulary that translates human intent into machine-optimized instructions. By engaging in a structured process, you move from simply asking questions to actively shaping the answers.
Expert Insight: Cognitive Load
Consider how explicitly defining roles and constraints reduces the LLM's "cognitive burden." By narrowing the search space for the model, you enable it to allocate more attention to accurate reasoning rather than guessing your intent.
The Shared Vocabulary (Toolkit)
TPS is built on modular components — tools that fine-tune interactions in specific ways. These three components form the core of the TPS Toolkit.
- RD Role Definition: Assigning an "expert" persona (e.g., "Shakespearean Poet") to guide the AI’s communication style and depth.
- CD Constraint Definition: Setting boundaries (e.g., "Exactly 100 words") to ensure the output meets technical requirements.
- CP Checkpoint Definition: Creating structured pauses to review progress and realign before proceeding with long-running tasks.
Pro Tip: Layering
The most significant improvement comes from combining RD and CD. Forcing the LLM to operate within an expert persona while adhering to strict constraints dramatically increases the signal-to-noise ratio in the response.
Understanding the Tiers
TPS matches your communication structure to your goal. Higher tiers offer more precision but require greater upfront effort.
Tier 1: The Simple Request 🔨
For straightforward info retrieval or quick content generation. Direct instructions with optional context. Analogy: Hammering a single nail.
Tier 2: The Blueprint 📐
For tasks requiring specific tone, style, or format. Involves defining roles, outlining steps, and setting constraints. Analogy: Building a house from a blueprint.
Tier 3: The Brainstorming Session 🧠
For highly ambiguous or complex problems where both parties iterate to uncovering new possibilities. Analogy: Whiteboard session with a creative partner.
Practical Scenarios
Choosing the right tier determines how much guidance you provide the AI. Consider these examples:
Bakery Promotion (Social Media)
Tier 2: Use RD ("Social Media Manager") + CD ("Under 100 characters", "Include #LocalBaker").
Why? You need specific marketing tone and character limits that a simple request won't guarantee.
Scientific Explanation
Tier 2: Use RD ("Physics Professor") + CD ("No jargon", "Use analogies for a child").
Why? A Tier 1 request ("Explain quantum physics") often results in technical overload that misses the audience level.
Role Definition (RD) Deep Dive
A strong RD isn't just a profession; it’s a persona built on four pillars of context:
Next Steps
You've completed the foundational module. You now understand the tools (RD, CD, CP) and the tiers of interaction.
Mastery Requirements:
- ✓ Document Success: Keep a record of RD/CD pairs that produce excellent results.
- ✓ Apply the Methodology: Before prompting, ask: "What level of structure does this task actually require?"
Ready to Transform Your AI Interactions?
Let's discuss how systematic interaction design can improve your organization's AI outcomes.
Start a Conversation