localhost:<insert-city>

Home screen
You can plan trips to San Francisco, UK and Argentina.
Conversation screen

Team

Tom Piaggio (Argentina) - Elevenlabs and backend.
Gareth Haagman (England) - Frontend, Google Places integration and backend work.
Eugenio Scafati (Argentina) - Frontend, Yelp integration and Vercel deployment.

Inspiration

When we first met at the hackathon, we realised that we were all visiting San Francisco for the first time. So the first thing we asked each other was what events each other were going to. We agreed that it was tough to understand where we should stay, work, go and eat without having any local guide us on the process.

It is currently too difficult to plan a trip without laborious As an illustrative example, Tom and Eugenio actually had to leave yesterday early from the Hackathon to meet with someone in Palo Alto because of our poor scheduling skills and Gareth wasn't capable of assisting a cool Groq event because he discovered while here that it was far away from where he is staying.

We are building the tool we wished we had when organizing our trip to San Francisco.

What it does

LocalHost is a voice agent that gathers information about places or events to go for work, hack, meet people and hangout from different sources like Luma, Google Places, Yelp and the web. It collates the information and sends a final itinerary report to the user's email.

How we built it

Our frontend was built with Lovable and deployed to Vercel. Our backend is a a lightweight Express app which uses ElevenLabs conversational agent and custom tools we built for the ElevenLabs agent such as Luma, Yelp, Google Places, and Perplexity for web search.

Frontend repo: https://github.com/scafati98/LocalHost-UI
Backend repo: https://github.com/tomaspiaggio/LocalHost

Challenges we ran into

Luma doesn't have an API to gather public information so we hacked our way through it.
Elevenlabs doesn’t have an autosave so we accidentally deleted everything.
Elevenlabs doesn't has a team collaboration feature (or we didn't find it).
Vercel free tier doesn't allows multiple contributors, so only one of us could push code to the UI.

Accomplishments that we're proud of

We collaborated well with each other and found a common problem we had all experienced.
We managed to build something that we will use ourselves.

What we learned

Lovable is great.

What's next for localhost:

We have the following list of upcoming features we would look to provide for our users

Amazon locker finder.
Google calendar integration, to produce an itinerary.
Apple Wallet / Google Pay integration to directly pay for events or reservations.
Note any demo requests you can apply for.
Integrate with weather data, to pick the best week to go to the city.
Calculate costs and time for each transportation alternative (public, uber/lyft, rental car).
Compare trip dates based on your objectives (when is it better to travel to X city when looking for Y results?)
Connect with people with similar objectives via social media.
Multi-person trips (i.e. add an integration to plan a trip for a group of friends/family with a shared itinerary).
Integrate Travel Insurance.
Check event dress codes and local weather to decide what clothes to pack.
Co-working or coffee places for work.
Index our own data from the web to reduce latency from Perplexity (3-5 seconds).

Business Model

Free for regular users. Google Ads integration and companies can pay for sponsored Events/Hotels/etc.
Enterprise: Paid subscription for corporate travel planning, integrations with local conferences, business travel and all of this can automatically be reconciled with SAP concur for expense management.

Our Moat

As we index our own data about consumer behavior, trends, seasonality, characteristics about different locations, we can provide lower latency responses that are much more personalized, we can improve recommendations with reinforcement learning (by seeing what events people actually go to).