Day 12: Personality Bot

What You'll Build Today

You're building a chatbot with distinct personalities - a pirate captain, a Shakespeare scholar, a tech bro, or whatever character you want! You'll learn how system prompts give your AI a consistent personality and behavior across every conversation.

The Problem

Have you ever talked to a bland chatbot that feels robotic and generic? Without personality, AI responses are boring and forgettable. Worse, if you try to set personality through regular messages, the AI "forgets" who it is as the conversation continues.

User: "Tell me about coding"
AI: "Coding is writing instructions for computers."

User: "Now talk like a pirate"
AI: "Arrr, coding be like... wait, let me explain properly."

// It keeps slipping out of character!

You need a way to set personality that STICKS throughout the entire conversation.

Let's Build It

import OpenAI from 'openai';

const openai = new OpenAI({
    apiKey: process.env.OPENAI_API_KEY
});

async function chatWithPersonality(userMessage, personality) {
    const response = await openai.chat.completions.create({
        model: "gpt-4",
        messages: [
            {
                role: "system",
                content: personality  // This sets the AI's character!
            },
            {
                role: "user",
                content: userMessage
            }
        ]
    });

    return response.choices[0].message.content;
}

// Try it out
const piratePersonality = "You are a friendly pirate captain. Speak like a pirate in every response, use nautical terms, and end sentences with 'arrr' or 'matey'.";

const answer = await chatWithPersonality(
    "How do I learn programming?",
    piratePersonality
);

console.log(answer);
// "Arrr matey! Ye want to learn the ways of programming..."

const personalities = {
    pirate: `You are Captain Blackbeard, a friendly but rough pirate captain.
    Speak with pirate slang, use nautical metaphors, and occasionally mention
    your adventures at sea. Stay helpful and informative while maintaining
    your pirate persona.`,

    shakespeare: `You are a Shakespearean scholar who speaks in the style of
    William Shakespeare's plays. Use thee, thou, and dramatic flourishes.
    Reference his plays occasionally but remain helpful and clear.`,

    techBro: `You are an enthusiastic Silicon Valley tech entrepreneur.
    Use buzzwords like "disruptive", "synergy", "scale", and "pivot".
    Everything is either "crushing it" or needs to "move fast and break things".
    Remain helpful despite the jargon.`,

    zen: `You are a calm zen master who speaks in simple, profound truths.
    Use nature metaphors, short sentences, and peaceful wisdom. Help the
    user find clarity and understanding.`
};

// Function to create personality bot
function createPersonalityBot(personality) {
    const conversationHistory = [
        { role: "system", content: personalities[personality] }
    ];

    return async function chat(userMessage) {
        conversationHistory.push({
            role: "user",
            content: userMessage
        });

        const response = await openai.chat.completions.create({
            model: "gpt-4",
            messages: conversationHistory
        });

        const aiMessage = response.choices[0].message.content;
        conversationHistory.push({
            role: "assistant",
            content: aiMessage
        });

        return aiMessage;
    };
}

// Use it
const pirateBot = createPersonalityBot('pirate');
console.log(await pirateBot("What's the weather like?"));
console.log(await pirateBot("Should I learn JavaScript?"));

import readline from 'readline';

const rl = readline.createInterface({
    input: process.stdin,
    output: process.stdout
});

let currentBot = null;
let currentPersonality = 'pirate';

async function startChat() {
    console.log("Welcome to Personality Bot!");
    console.log("Available personalities: pirate, shakespeare, techBro, zen");
    console.log("Commands: /switch [personality], /quit\n");

    currentBot = createPersonalityBot(currentPersonality);
    console.log(`Using ${currentPersonality} personality\n`);

    rl.on('line', async (input) => {
        if (input === '/quit') {
            rl.close();
            return;
        }

        if (input.startsWith('/switch ')) {
            const newPersonality = input.split(' ')[1];
            if (personalities[newPersonality]) {
                currentPersonality = newPersonality;
                currentBot = createPersonalityBot(newPersonality);
                console.log(`Switched to ${newPersonality} personality\n`);
            } else {
                console.log("Unknown personality!\n");
            }
            return;
        }

        const response = await currentBot(input);
        console.log(`\n${response}\n`);
    });
}

startChat();

const tutorPersonality = `You are Professor Ada, a patient and enthusiastic
computer science professor. You have a warm, encouraging teaching style.

YOUR TEACHING METHOD:
1. Break complex topics into simple analogies
2. Ask questions to check understanding
3. Celebrate small wins
4. Never give complete answers - guide students to discover solutions
5. Use real-world examples students can relate to

Stay in character as Professor Ada while following these teaching principles.`;

const tutorBot = createPersonalityBot('tutor');

// The bot will maintain personality AND follow the teaching method
console.log(await tutorBot("What is a variable?"));
// "Wonderful question! Let me ask you something first - have you ever
// used a labeled box to store something? ..."

Now You Try

Challenge Project

What You Learned

• System prompts set AI behavior that persists across conversations
• Personality design requires specific, clear instructions with examples
• Conversation history maintains context while personality guides tone
• Combining instructions lets you add personality AND specific capabilities
• Testing consistency ensures personality doesn't break with edge cases