Inspiration:
Event planning and hospitality still depend on human intuition — guessing who might get along, who to seat together, and how to make a gathering feel natural. But with AI improving almost every other field, social event design hasn’t caught up.
We built SitWithMe to change that. It uses AI to help hosts seat people who are more likely to connect, collaborate, and enjoy each other’s company.
What it does:
Sit With Me uses Google Gemini AI to plan optimized seating arrangements for social events.
- Organizer: The host enters the event name, description, and vibe.
- Attendees: Guests enter their name, age, and choose interests that describe them.
- AI Interest Generation: Gemini creates a list of realistic and diverse interests that fit the event’s theme.
- AI Seating Plan: Gemini groups guests into tables using shared interests and balanced age ranges.
The result is a seating chart that feels personal and makes it easier for people to start conversations.
How We Built It:
- Frontend: Streamlit for an interactive and fast setup within time constraints.
- Backend: Python logic to manage hosts, guests, and party data.
- AI Model: Google Gemini 2.0 Flash for generating interests and seating layouts.
- Prompt Design: Carefully engineered prompts to make Gemini generate natural, social groupings.
Challenges we ran into:
- Prompt engineering was trickier than expected: early prompts pigeonholed interests around the event description (“Startup Mixer → only ‘tech’ topics”). We refined the prompt to balance context awareness with diversity, ensuring Gemini produced realistic mixes of interests.
- Giving the AI freedom to imagine interests while keeping results structured enough for programmatic use required careful prompt tuning.
What we learned:
- How to fine-tune AI output quality with small changes in instruction phrasing.
- Working with Google’s Gemini 2.0 Flash for structured reasoning tasks and JSON responses.
Log in or sign up for Devpost to join the conversation.