Free Course from Omagine LabsOmagine Labs
Intro1The Prediction MachineJagged IntelligenceProbabilistic vs. DeterministicBe More SpecificAdvanced TechniquesThe Head StartHallucinationsMemoryThe Human in the LoopThe AI LandscapeAI JargonWhat's Next0/12
05
LEVEL UP⭐⭐

Advanced Techniques

RACE, CLEAR, RISE and other frameworks

When your first output from AI isn’t great, most people either give up or retype the same vague prompt. This lesson gives you a toolkit for iteration: specific levers you can pull to shape the output. These aren’t rules to memorize — they’re ways of thinking that help you debug bad results.

Three frameworks, one goal

Each framework is just a different way of organizing the same ingredients you learned in the last lesson. Pick the one that clicks with how your brain works — or ignore them all and just make sure you’re covering the bases.

Role, Action, Context, Expectation

R

Role

Who should the AI pretend to be? A senior strategist? An editor? A coach?

A

Action

What specific action do you want? Draft, analyze, brainstorm, critique?

C

Context

What background information does the AI need to do this well?

E

Expectation

What does the ideal output look like? Format, length, tone, quality bar?

Example Prompt
Role: You are a senior business strategist who specializes in executive communication.

Action: Create a set of talking points for a quarterly business review presentation.

Context: I lead a product team of 12 at a mid-size SaaS company. This quarter we launched two features, missed one revenue target, and grew our user base by 15%. The audience is my VP and two C-suite executives.

Expectation: Provide 5-7 concise talking points that balance honesty about the miss with confidence in the trajectory. Tone should be direct and data-driven.

Which one should you use?

All three work. They just emphasize different things — RACE is role-heavy, CLEAR is format-heavy, RISE is process-heavy. Pick whichever fits how your brain naturally works. Or ignore the frameworks entirely and just make sure your prompts include the key ingredients: role, context, examples, format, and constraints. The frameworks are training wheels, not permanent fixtures.

When an output isn’t right

Before you retype or start over, run through these four diagnostic questions. Usually, the fix is adding one missing ingredient.

01

Did I give it a role? (Who should the AI be?)

02

Did I specify the format? (Bullets, prose, table, length?)

03

Did I include an example of what good looks like?

04

Did I give it enough context about the situation?

Takeaway: The point is not to memorize a framework. The point is to internalize the ingredients: role, context, examples, format, constraints. When an output isn’t right, think about which ingredient is missing. That’s your debugging superpower.

One of the fastest-growing ways people interact with AI is by talking to it. If typing out detailed prompts feels like work, try using speech-to-text — the built-in voice features in ChatGPT and Claude’s mobile apps, or your device’s native dictation.

Now that you know the ingredients of a good prompt, it’s often easier to say them out loud than to type them. “Hey, act as a senior marketing strategist and help me write a campaign brief for our Q2 launch. The audience is our VP of Marketing. Keep it to one page, bullet format.” That took ten seconds to say and would have taken a minute to type.

Voice input is especially powerful for the context-heavy prompts we’ve been building. You already know how to brief someone verbally — you do it in meetings every day. Let that same skill carry over to AI.

Need help bringing this to your team?

We work alongside your team to build AI-native workflows — from one-week sprints to full engineering acceleration. No handoffs, no slide decks.

Talk to us