Persona Generator and Chat - Quick Start Guide

Create a persona in the format of Nemotron Personas with ChatGPT, modify it / iterate on it, save it, and chat with it.

🖼️ What It Looks Like

Main Interface

Clean, intuitive navigation between persona generation and chat features.

🧠 Persona Generator

Create detailed personas using natural language prompts, then iterate and refine them.

💬 Persona Chat

Have realistic conversations with your saved personas for market research and testing.

---

🚀 Quick Setup

Step 1: Get Your API Keys

OpenAI API Key

1. Go to OpenAI's website
2. Sign up or log in
3. Create a new API key
4. Copy the key (starts with sk-...)

Hugging Face Token

1. Go to Hugging Face tokens page
2. Sign up or log in
3. Click "New token"
4. Choose "Read" permission (sufficient for this app)
5. Copy the token (starts with hf_...)

Step 2: Set Up Your API Keys

Create a file called .env in this folder and add:

OPENAI_API_KEY=your_openai_key_here
HF_TOKEN=your_hf_token_here

(Replace with your actual keys)

Step 3: Start the App

Choose your method:

🖱️ Easiest: Double-Click Method

• Windows: Double-click start_personas.bat
• Mac:
  • Option 1: Run python create_mac_app.py then double-click "Persona System.app"
  • Option 2: Double-click start_personas.py (if Python is set as default)
  • Option 3: Open Terminal and run bash start_personas.sh
• Linux: Double-click start_personas.sh
• Any OS: Run python start_personas.py

Note you do not need to give your email to streamlit when it prompts you with If you'd like to receive helpful onboarding emails, news, offers, promotions, and the occasional swag, please enter your email address below. Otherwise, leave this field blank.

💻 Command Line (if you prefer)

cd frontend
pip install -r requirements.txt
streamlit run app.py

🎉 That's It!

Your browser should open automatically to http://localhost:8501

If not, try to open up that in your browser. On the first time you run it, it may take a minute or two to download the NemoTron personas dataset, and each time you try to create a new persona there will be some latency to wait for OpenAI to generate a new persona for you.

🆘 Troubleshooting

"Python not found"

• Install Python from python.org
• Make sure to check "Add to PATH" during installation

"Module not found" errors

The launcher scripts will automatically install required packages, but if you get errors:

pip install -r frontend/requirements.txt

API Key Issues

• Make sure your .env file is in the main project folder (same level as this README)
• Check that your API key is correct and has credits
• The key should start with sk-

Port Already in Use

If you get a port error, try:

streamlit run app.py --server.port 8502

🎭 Complete Workflow: From Creation to Chat

Phase 1: Creating Your First Persona 🧠

1. Start the App: Use any method above to launch the system
2. Navigate to Persona Generator: Click "🧠 Persona Generator" from the home page
3. Choose Your Model: Select from available models (gpt-5, gpt4o, etc..)
4. Create Initial Persona: Enter a prompt describing a high-level persona in your target audience / target stakeholder group:

   Generate a persona of a procurement decision maker at a major retailer like Costco
   

5. Review the Generated Persona: The system creates a detailed persona with:
   • Core personality traits
   • Professional background
   • Sports/fitness interests
   • Arts and culture preferences
   • Travel style
   • Culinary tastes
   • Skills and expertise
   • Career goals and ambitions
   • Structured lists of interests and skills

Phase 2: Iterating and Refining 🔄

6. Iterate on Your Persona: After the first generation, the prompt changes to allow iteration:

   Make them more 10 years older and more conservative in their outloook.
   

   Or create variations:

   Keep the marketing background and make them more in touch with the Gen-Z demographic 
   

7. Multiple Iterations: Keep refining until you get one or more personas you are happy with.

Phase 3: Saving Your Persona 💾

9. Save When Ready: Once you're happy with a persona, use the save section:

   • Give it a name: you will load it later for the chat feature
   • You can leave blank for auto-generated name
   • Click "💾 Save Persona"

10. Confirmation: You'll see "✅ Persona saved as: [name]" and guidance to find it in Persona Chat

11. Note on Generating Multiple Personas The persona generator only remembers the most recently proposed persona. If you like it, save it. Note as well that creating more personas is not mutually exclusive: after you save a persona, you can continue to ask ChatGPT to continue iterating on it until you get a new persona that you are happy with.

Phase 4: Chatting with Your Persona 💬

11. Navigate to Persona Chat: Click "💬 Persona Chat" from the home page

12. Select Your Persona: Choose from the dropdown of saved personas

13. Configure Chat Settings:

    • Choose your preferred model
    • Adjust temperature for creativity level

14. Start Chatting: Begin conversations however you like!

15. Realistic Responses: The persona is instructed to respond in character, drawing from their detailed background. They are also instructed to be realistic, but of course take it with a grain of salt.

🎯 Example Complete Workflow

Goal: Create a persona for testing a new fitness app

1. Generate: "Create a busy working parent interested in fitness but struggling with time"
2. Iterate: "Make them more tech-savvy and add experience with fitness apps"
3. Refine: "Add that they prefer home workouts due to childcare constraints"
4. Save: Name it "Busy_Parent_Fitness"
5. Chat: Test questions about the fitness app:
   • "What's your biggest challenge with staying fit?"
   • "How do you currently track your workouts?"
   • "What would make you choose one fitness app over another?"

Need Help? Try to create an issue on GitHub.