AI Persona Prompting

Introduction

Persona prompting assigns the AI a specific role — 'You are a senior backend engineer' or 'You are a friendly nutrition coach'. Personas dramatically shift tone, depth and even the model's accuracy on specialised tasks.

Beginner analogy: Asking your friend who happens to be a doctor for medical info gets a different answer than asking your friend who is a poet — even if they read the same Wikipedia page.

Understanding the topic

Core concepts to understand:

• Set the role explicitly: 'You are a senior tax advisor in the UK.'
• Add expertise level (junior, senior, world-class).
• Add tone (friendly, formal, concise).
• Combine with output-format rules for full control.

Syntax reference

Visual workflow / architecture:

Same question, different personas:

  "Explain TLS handshakes."
        │
   ┌────┴───────┐
   ▼            ▼
[5-yr-old]   [security engineer]
"Like a      "Client sends ClientHello
secret       with cipher suites and
handshake    randomness, server replies
between two  with cert chain, ECDHE
friends."    key exchange, ..."

Real-world use

Khan Academy's Khanmigo uses persona prompting to adapt to grade level. Replika and Character.AI are entire products built on persona prompts.

Best practices

• Be specific: 'senior staff engineer at Google' beats 'expert'.
• Combine persona with output format and tone rules.

Hands-on exercise

Interview preparation — practice these questions:

• Q1. What is persona prompting?
• Q2. Why does adding a role improve answers?
• Q3. How would you persona-prompt a financial-advice bot?