Touch Grass

Touch Grass helps people find and join local activities by automating event creation and scheduling when enough nearby people express the same interest.

Comment

Inspiration

Starting at a new campus like Georgia Tech can feel overwhelming especially for freshmen and exchange students who could struggle to find community. Our team is made up of 2 exchange students as well and we are grateful for all the events and opportunities to go out and meet new people. But, we realise that is only possible because we live on campus and in a living-learning community. Hence, we wanted to expand this possibility and make it easier (and a little funnier) to literally “Touch Grass”: step outside, connect, and do things together instead of staying isolated in a new environment.

What it does

Touch Grass is an AI-powered event concierge that does the hard part of social planning for you.

  1. Sign up & set your profile: choose your comfort levels, tags, and interests.
  2. Tell the AI what you want: “I want to watch the football game.”
  3. AI becomes the event planner: instead of just finding existing events, the AI actually creates new ones by matching people with similar interests, proposing times/venues, and checking acceptance criteria (group size, gender ratio, cost). Perfect for students who are shy or don’t want to be the one to initiate.
  4. Smart confirmation: one day before, the app confirms or cancels based on turnout, then shares contact info and carpool suggestions. The result: Students end up at the right events, with the right people, at the right time — without the stress of planning or awkwardly asking “who wants to hang out?”

How we built it

  • Frontend in React + TypeScript using Vite and Tailwind for fast iteration and styling.
  • Backend in Node.js with Express, connected to a PostgreSQL database.
  • AI event planner built on top of the Gemini API, with additional heuristics to handle scheduling rules and conflict checks.
  • Containerized PostgreSQL in Docker to keep setup simple and reproducible.
  • Used GitHub for version control and collaborated with a shared design in Figma to align on UI/UX.

Challenges we ran into

  • Getting PostgreSQL running consistently on different machines (especially Windows vs. Mac) and making sure the database URL worked.
  • Integrating the Gemini API in a way that produces structured, reliable JSON output instead of freeform text.
  • Avoiding duplicate or conflicting events when multiple people had similar interests.
  • Balancing AI creativity with clear rules so that events were realistic, safe, and actionable.

Accomplishments that we're proud of

  • A fully working demo where students can create profiles, add interests, and get auto-matched to events.
  • Clean UI that feels calming and approachable for new students.
  • Event confirmation workflow (with cancellation logic) that makes the app practical.
  • Incorporating Gemini API to generate event ideas and descriptions, then grounding them with real venues and schedules.

What we learned

  • The importance of think of the user flow first before diving into code. By mapping out the journey - from sign-up, to telling the AI what you want, to event confirmation - we could align design, copy, and backend logic around the same experience.
  • The need for safety features in building community tools (filters, cancellation rules, privacy).
  • AI is powerful, but structured heuristics (rules) make it more reliable in real use cases.
  • Relational databases are powerful tools to organize data, but require unique considerations particularly when designing many-to-many relationships

This made the whole project more cohesive: the UI feels intentional, the AI features make sense, and the app solves a real pain point instead of just being a collection of features.

What's next for Touch Grass

  • Mobile-friendly version so students can quickly RSVP and get reminders on the go.
  • Deeper integration with campus calendars and local venues to automatically suggest available times and spaces.
  • More robust group chat and carpooling features to make coordination even easier.
  • Additional safety features like verified hosts and optional ID checks for certain events.
  • Scaling beyond campus: adapting the app for neighborhoods, hobby groups, or city-level meetups.
Share this project:

Updates