Partybank: Find events that match your mood

Inspiration

PartyBank was built to answer a question we’ve all faced: What should I do tonight, based on how I actually feel? Instead of filtering by genre or location, we wanted to build an AI-powered system that starts with human emotion and ends with real-world connection.

What it does

PartyBank is an AI-powered mood-to-event matchmaker. Users describe how they’re feeling in natural language, and the app:

• Analyzes their mood using an LLM via OpenRouter, targeting deepseek:free.
• Finds local events using Tavily’s API in both background and foreground agent modes.
• Checks public sentiment about those events using Bright Data’s API.
• Returns recommendations that either match or improve the user’s emotional state, with real-time vibe checks from the web.

How we built it

We developed the project in Replit, using OpenRouter to power LLM-based mood extraction from user prompts and sentiment analysis of live web data.

Event discovery is handled through Tavily. It operates in a background mode every 24 hours to gather new new events into an internal database. Search queries will return results from the local database quickly but also run a new search in Tavily to get the most relevant and up to date results.

Public opinion on events is gathered using Bright Data. It does a SERP API request against twitter (through Google) and another against reddit (also through Google). These search results are fed into an LLM through OpenRouter to get a sentiment analysis.

Conclusion

Our biggest challenge was building a pipeline that could turn something as abstract as emotion into a concrete, trustworthy recommendation but this is a great use-case for AI. PartyBank is more than just event search—it’s a step toward AI that supports emotional wellness and informed social decisions.

Built With

• brightdata
• deepseek
• openrouter
• replit
• tavily