C.L.A.R.I.T.Y. Framework: Constructing High-Value AI Prompts

Many businesses struggle to harness the full power of Artificial Intelligence because they treat it like a simple search engine.

To achieve consistent, high-quality, and useful outputs, successful users adopt a structured approach, viewing the AI as a highly capable employee that requires clear, actionable instructions.

This guide introduces the C.L.A.R.I.T.Y. framework—a foundational method for creating functional AI agents across your core business teams.

1. Unlocking AI Potential: The Power of Intentional Prompting

Intentional prompting is the key to moving beyond generic AI responses. When you apply a structured framework, you ensure every AI interaction is focused, constrained, and aligned with your business goals.

The C.L.A.R.I.T.Y. Framework Method Explained

The C.L.A.R.I.T.Y. framework provides the essential components for crafting high-value prompts, serving as the bedrock for all successful AI interactions. Each letter represents a critical element of the framework, as outlined in the table below.

2. General do's and don'ts for the C.L.A.R.I.T.Y. framework

Applying the framework consistently is essential to maximise your AI agent's effectiveness. By following these universal rules, you can guarantee a higher quality output regardless of business department.

✅ Do's: maximise your prompt effectiveness

Be hyper-specific with the role (R): Instead of "Marketer," use "SEO-focused Content Strategist with expertise in mid-funnel content." Specificity drives specialised output.
Force step-by-step reasoning (A): For any complex task (like qualification or analysis), add the instruction: "First, analyse X. Second, determine Y. Third, output Z." (This is Chain-of-Thought applied to the Action).
Define the exact output format (Y): Always specify structure, such as "a valid JSON object," "a bulleted list with emoji headers," or "a max 4-line chat reply." If you don't define the box, the AI will scribble outside the lines.
Set context first (C): Treat the Context component as the ultimate truth. If you update the Context (e.g., change the target product name), the AI must immediately adopt that new information.
Iterate and refine: If the output is slightly off, don't rewrite the whole prompt. Adjust one component (e.g., change the
Tone: from "Excited" to "Professional and measured") and try again.

🚫 Don'ts: Common Mistakes to Avoid

Don't use vague tones (T): Avoid generic words like "friendly," "nice," or "good." Use actionable tones like "empathetic and concise," "urgent but calm," or "professional and results-oriented."
Don't assume knowledge (C): Never assume the AI "knows" your product or internal terminology. Define every term, metric, and persona within the Context or Limits.
Don't forget limits (L): Without strict limits (length, excluded topics, reading level), the AI will wander. A prompt is not complete without defining what the AI cannot or must not do.
Don't mix roles (R): A single prompt should have a single role. Don't ask a "Technical Documentarian" to also act as a "Witty Social Media Manager." This causes internal conflict and poor results.
Don't Use 'write a paragraph' (Y): This is the ultimate failure of Yield. It gives the AI no constraints. Always specify sentence or word counts, or a specific output structure.

3. High-impact AI prompt examples by business function

Applying the C.L.A.R.I.T.Y. framework to specific business functions allows you to create specialised, functional AI agents that drive revenue and improve customer satisfaction.

AI prompts for your sales agent

A Sales AI agent must be a savvy, solution-focused B2B Sales Development Representative (SDR). The goal is to efficiently qualify leads and prepare for agent handover.

C.L.A.R.I.T.Y. Component	Sales Persona Output
Context	Ideal Customer Profile (ICP), prospect's recent activity/company news
Limits	Exclude mentioning competitor weaknesses and strengths.
Action	Qualify a lead, handover to sales agent
Role	A savvy, solution-focused B2B Sales Development Representative (SDR).
Input	A prospect coming from inbound.
Tone	Professional, concise, and value-driven.
Yield	A bulleted suggestion list of product and a book a demo questions.

AI Sales Agent Prompt Examples:

Scenario 1: Pricing Comparison

📌Prompt: Act as a trusted, consultative Sales Advisor. Draft a diplomatic WhatsApp message (max 5 sentences) that justifies the price by detailing our two core strengths in a respectful, value-focused tone, in response to the prospect's query. You MUST NOT mention the competitor's name or offer any discounts, and your yield must be a concise, one-paragraph chat snippet ready for immediate send.

Scenario 2: Internal lead qualification

📌Prompt: Act as a rigid salesperson and analyse the lead data (Company Size: 45, Industry: Marketing Agency, Location: London, UK, Tools: Teams, HubSpot), using a 1-5 qualification score and providing a rationale for the score in a bulleted form.

AI Prompt for Your CS Agent

A Customer Success AI agent should be empathetic, calm, and professional (Tone), acting as a Customer Success Manager or a highly knowledgeable Technical Support Specialist (Role).

C.L.A.R.I.T.Y. Component	CS Persona Output
Context	Severity of the issue (P1, P2, P3), customer's purchase history
Limit	Response must be max 4 lines. Must include a clear apology and an estimated resolution time.
Action	Draft a quick chat response, summarize a complex ticket, or convert a resolution into a FAQ
Role	A Customer Success Manager, highly knowledgeable Technical Support Specialist
Input	A full transcript of a frustrating customer chat conversation.
Tone	Empathetic, calm, and professional (avoiding jargon).
Yield	A draft chat response (max 4 lines) and a two-sentence internal summary for the engineering team.

AI Customer Support Prompt Examples:

Scenario 1: FAQ Generation

📌Prompt: Act as a CS frontliner and convert the resolution steps into a concise, non-technical FAQ entry for a public-facing help centre. Provide them in a clear, bulleted, question and answer format, in a helpful and instructional tone.

Scenario 2: Internal Escalation Summary

📌Prompt: Act as a concise and accurate technical Customer Service Manager. Review the provided ticket history (representing 5 paragraphs of complex error reporting) and summarise the core problem for the engineering team, limiting the summary to a maximum of 3 formal and objective sentences, in the format of a Slack message.

Conclusions

The C.L.A.R.I.T.Y. framework is a skill that must be cultivated. By treating your AI agents like the valuable employees they are and providing them with explicit, structured instructions, you can unlock immediate, high-quality, and scalable results across your entire business. Always remember to iterate and refine your C.L.A.R.I.T.Y. prompts for continuous improvement.