Core Framework: TCREI

(Tiny Crabs Ride Enormous Iguanas)

1. Task - What you want the AI to do
2. Context - Background details (more = better output)
3. References - Examples to clarify what you want
4. Evaluate - Is this output what I need?
5. Iterate - Refine until satisfied (ABI: Always Be Iterating)

4 Iteration Methods: RSTI

(Ramen Saves Tragic Idiots)

1. Revisit framework - Add more context, references, or persona
2. Separate sentences - Break complex prompts into simpler chunks
3. Try different phrasing - Use analogous tasks (marketing plan → compelling story)
4. Introduce constraints - Narrow focus (specific region/tempo/theme)

Enhance Any Task With:

• Persona: Role for AI to embody ("Act as an anime expert...")
• Format: Structure of output (table, bullet points, paragraphs)

Multimodal Prompting

Works with images, audio, video, code - same TCREI framework applies

• Specify input/output modalities clearly
• Provide context about the media you're using

Critical Safeguards

Hallucinations - AI makes up incorrect information Biases - Human biases reflected in outputs Solution: Human-in-the-loop - always verify outputs yourself

Advanced Techniques

1. Prompt Chaining

Connect prompts in sequence, each building on previous output

• Example: Generate summary → refine tagline → create marketing plan

2. Chain of Thought

Add "Explain your thought process" to prompts

• Helps understand AI reasoning
• Improves decision-making quality

3. Tree of Thought

Explore multiple reasoning paths simultaneously

• Prompt: "Imagine three different experts pitching ideas..."
• Useful for complex/abstract problems
• Combine with Chain of Thought for feedback at each step

4. Meta-Prompting

Use AI to help create better prompts when stuck

Follow @godofprompt for more AI tips.