SentiMuse

• 

  SentiMuse: The Prompt Whisperer for Emotional Resonance

Inspiration

We wanted to build something that felt deeply human, a tool that speaks the emotional language of creators and helps them express feelings, not just ideas. The challenge was to make generative AI feel less like a machine and more like a mindful collaborator.

What SentiMuse does

SentiMuse co-creates emotionally aware prompts based on your mood, topic, and optional journaling notes. It uses sentiment detection + human-first prompt generation to create writing prompts you can use in ChatGPT, Claude, or Gemini.

How we built SentiMuse

• Built in Python using Streamlit for UI
• OpenAI’s GPT-3.5-turbo for prompt generation
• VADER for sentiment detection
• ReportLab for PDF export
• Emotions were manually tagged with emoji + color
• Custom logic and feedback system with session state

Challenges we ran into

• Designing prompts that feel emotionally nuanced and not robotic
• Avoiding poetic/metaphorical fluff that LLMs usually generate
• Ensuring the user experience was both reflective and easy-to-use
• Maintaining emotional consistency across UI and prompt tone

Accomplishments that we're proud of

• Building a fully functional emotional prompt app in 48 hours
• Creating a genuinely useful tool for creators and mental health contexts

What we learned

• Prompt engineering for emotional tone is hard, but possible
• Building with empathy leads to more meaningful AI interactions
• Streamlit is amazing for rapid prototyping + UI

What's next for SentiMuse

• Enable multi-turn prompt co-creation
• Integrate offline mode using ONNX models
• Launch a mobile-friendly version

Built With

• openai
• python
• reportlab
• streamlit
• vader