Inspiration
The inspiration for BridgePath comes from a simple but devastating reality: Your address shouldn't determine your future.
Imagine two students with the exact same talent. Student A lives 15 minutes from a university, has a car, and knows about internships. Student B lives in a transit desert, doesn't know about scholarship deadlines, and can't afford the bus fare to get to an interview. Student B misses the opportunity. The difference isn't talent it's access.
Millions of people miss out on life-changing opportunities (jobs, scholarships, healthcare, training) not because they aren't qualified, but because they are blocked by information barriers, mobility barriers, and hidden logistical costs. We wanted to build something that doesn't just "summarize text," but actively removes the friction between a person and their future.
What it does
BridgePath is an access engine. It turns any opportunity document (an interview invitation, a scholarship letter, a government notice) into a concrete, real-world action plan.
Instead of just telling you what a document says, BridgePath follows a core loop: Opportunity → Understand → Reach → Act.
- Understand: It reads the document and explains it in plain language, highlighting exactly what is required.
- Reach: It calculates the best way to get there, optimizing not just for distance, but for the user's constraints (budget, time, safety).
- Act: It detects hidden barriers (e.g., "I need to print a resume but don't have a printer") and unlocks local assistance (e.g., "Free printing at the Central Library"), generating a step-by-step checklist and calendar reminders.
How I built it
I built BridgePath as a modern web application focused on a single, elegant workflow.
- Frontend: I used Next.js (App Router), React, and TypeScript to build a fast, responsive, and accessible UI. I styled it with Tailwind CSS to create a clean, dark-mode interface that is easy to read and reduces cognitive load for stressed users.
- AI Engine: I integrated the OpenAI API (using
gpt-4o-mini) with strict JSON schema prompting. This allows the app to reliably extract structured data (dates, locations, requirements, hidden support mentions) without hallucinating fake benefits. - Routing & Logic: I built a custom routing adapter. While the demo uses controlled mock data to guarantee the "wow" factor of the time-savings comparison, the architecture is designed to plug directly into GTFS (transit) and Google Maps APIs, overlaying social service data on top of geospatial routing.
Challenges we faced
- Scope Creep: My initial idea was to combine healthcare, legal, and education into one app. I quickly realized that "Swiss Army knife" apps lose their narrative. The biggest challenge was being ruthless and cutting features to focus on one coherent workflow: taking a document and turning it into a travel and action plan.
- AI Reliability: LLMs can be unpredictable. If the AI hallucinated a date or missed a requirement during a live demo, the app would break. I solved this by building a robust fallback system and using highly constrained system prompts to force structured JSON outputs.
- Contextual Integration: The hardest technical challenge was making the different features "talk" to each other. I didn't want a chatbot next to a map. I had to write logic so that the route planner knows about the interview time, and the task checklist knows about the library hours.
Accomplishments that I am proud of
I am incredibly proud of the "Route Comparison" UI. Seeing a daunting 2 hour 18 minute commute with 3 transfers shrink down to a 46 minute optimized route while simultaneously flagging that a transit fare voucher is available is a powerful emotional moment. It proves that the app isn't just processing data; it's giving people their time and dignity back.
What I learned
I learned that a single, elegant workflow beats a massive feature list every time. Hackathon judges (and real users) don't want an app that does 10 things poorly; they want an app that solves one massive problem beautifully. I also learned that the true value of AI isn't in generating text, but in contextual integration connecting information to physical action.
What's next for BridgePath
- Live Transit Integration: Connecting to real-time GTFS feeds and local transit authority APIs to provide live bus tracking and actual fare calculations.
- OCR & Mail Scanning: Adding computer vision so users can simply take a photo of a physical letter they received in the mail.
- Direct Service Partnerships: Partnering with local libraries and transit boards to allow users to claim the "assistance vouchers" directly through the app, rather than just telling them where to go.
Log in or sign up for Devpost to join the conversation.