AccessMate

Inspiration

I watched a streaming video of a blind influencer sharing her travel vlog. It was clear how challenging it was for her to find hotels with facilities that truly accommodate people with disabilities. It made me realize how much harder travel can be without accessible information. I believe we can use AI to build a tool that makes travel easier, safer, and more inclusive for everyone.

What it does

AccessMate is an AI-powered inclusive travel planner and companion. It helps people with disabilities, families with small children, or anyone with specific accessibility needs plan barrier-free trips. The system suggests destinations based on individual accessibility preferences (e.g., roll-in showers, quiet zones, sensory-friendly attractions), generates a personalized itinerary, and even drafts accommodation request emails or accessible transit directions using natural language AI.

How we built it

We used Databricks to build a robust data pipeline that ingests and processes data from public APIs (e.g., Booking.com). Cleaned and scored data is stored in Delta Lake tables, which feed into an itinerary generation engine that matches user preferences with suitable destinations and venues.

Offer conversational recommendations through a simple chatbot interface

Challenges we ran into

Databricks SQL Agent is not able to recognize nested column values for same cases, for example, array, JSON, etc.
Databricks SQL Agent doesn't work with views.

What we learned

Learnt how to build a chatbot to understand natural language and get translated to SQL for querying data.
Explored some AI features in Databricks, trying to make questions be answered more accurately.