Locolo

Great events go unnoticed daily, and Locoo makes it easy to find the perfect local, community-driven event that matches your tastes & preferences.

Comment

Inspiration

The inspiration behind our product emerged from a common frustration: despite the abundance of local, community-driven events and activities, we often find ourselves cycling through the same one or two activities with friends. We're always eager to discover new local spots and experiences, yet we frequently miss out on fantastic events that perfectly match our interests. The challenge lies in navigating multiple platforms to find events that not only capture our attention but also fit our unique tastes and preferences.

What it does

Locolo is a user-centric, AI-powered chat platform designed to simplify the search for local events: from events hosted by local businesses to those hosted by the city. By simply asking our ✨vibe-based✨ AI, "I want to have a romantic jazz date night" to "I want to party and listen to techno" users receive personalized, immediate recommendations, transforming how they explore and engage with their city's vibrant cultural landscape.

How we built it

We trained our model on hundreds of events happening this month. To match the user's input to an event, we created embeddings done through RAG. The embeddings were generated with hugging face’s "all-MiniLM-L6-v2."

Challenges we ran into

  1. We wanted to parse through Instagram accounts who post about local activities and events, but found that the LLM had a hard time figuring out whether a specific post is about an event or an image. We ended up using event APIs we were confident about that could give more accurate results.
  2. We wanted to create personas that would trigger different results that the user could choose from. However, limitations in the existing LlamaIndex chat interface made implementing personas a prohibitively difficult engineering task

What's next for Locolo

  1. Leverage meta data filtering for more accurate retrieval of events
  2. Explore additional event data sources for comprehensive coverage (i.e: partiful, Resident RA, meetup, fever, sofar sounds, Instagram / TikTok influencer pages, etc) 3.Add support for personas by augmenting prompts based on selected persona
  3. Implement user feedback mechanisms for continuous improvement.

Built With

Share this project:

Updates