Bonded — About the Project
Inspiration
College is one of the most important chapters of a person’s life, yet it remains surprisingly hard to navigate. Students are expected to make high-impact decisions academically, socially, and personally without a system that actually understands their environment.
The information students need does exist:
- what classes to take
- what clubs and organizations are active
- what events are happening
- who to ask for advice
But it’s scattered across emails, flyers, group chats, spreadsheets, and word-of-mouth. If you don’t already know where to look or who to ask you’re left guessing.
At the same time, we watched every major platform race to add “AI.” Instagram AI. Snapchat AI. TikTok AI. These tools respond to prompts, but they lack real-world context. They don’t know your schedule, your campus, your social boundaries, or your goals. Intelligence without context becomes noise.
Bonded was inspired by a simple question:
What if college had a system that actually understood students’ lives and guided them through it?
What It Does
Bonded is a context-aware campus system that helps students navigate college socially, academically, and emotionally.
At its core, Bonded combines:
- a verified campus social network
- a unified campus guide
- and a context-first AI assistant called Link
All working together as one system.
Students sign up using a .edu email and upload a screenshot of their class schedule. From that moment, Bonded understands their academic context.
Bonded provides:
- Yearbook — a directory of every verified student on campus
- Class group chats — automatically created for every enrolled course
- Forum — a single place for campus-wide discussion and advice
- Events feed — all campus events in one place
- Clubs & Organizations directory — searchable and verified
- Unified Calendar — classes and events combined into one weekly view
On top of this foundation is Link, an AI system that doesn’t just answer questions—it guides students using verified data, human input, and consent-driven connections.
How We Built It
Bonded was built as a full-stack, production-oriented system.
Core Architecture
- Frontend: React Native (mobile-first experience)
- Backend: Supabase (PostgreSQL, Row Level Security, real-time subscriptions)
- Authentication:
.eduemail verification - Messaging: real-time group and direct messaging
- Data modeling: structured representations of classes, schedules, clubs, events, and users
Link: System-Driven AI
Link is not a chatbot or GPT wrapper. Every message sent to Link passes through a deterministic pipeline:
- Intent detection — determines what the user is trying to do
- State management — enforces conversational and workflow rules
- Context retrieval — pulls verified campus and user data
- Confidence checks — prevents guessing and hallucinations
- Human outreach (if needed) — gathers real responses from students
- Consent workflows — ensures safe, opt-in connections
- Memory — stores only scoped, user-approved context
The AI never decides what action to take.
It only helps communicate after the system determines the correct, safe path forward.
We refer to this architecture as Linkya: a context-first, human-in-the-loop intelligence system.
Challenges We Ran Into
Modeling Real-World Context
College life is messy. Encoding schedules, classes, clubs, events, and social relationships into a clean, reliable system required careful schema design and strict validation.
Preventing AI Hallucinations
We intentionally designed Link to refuse to answer when confidence is low. This meant building confidence thresholds, verification rules, and fallback paths—something most AI systems skip.
Privacy and Consent
Connecting students safely required explicit opt-in workflows at every step. Cold DMs were not acceptable. Trust had to be enforced at the system level.
Scope Control
Bonded feels large because it’s integrated. Keeping every feature aligned with a single goal—helping students navigate college—was critical to avoid feature creep.
Accomplishments That We’re Proud Of
- Built a context-aware social system, not just an app
- Designed an AI architecture that avoids hallucinations by design
- Unified social, academic, and event data into one experience
- Implemented human-in-the-loop workflows for real-world questions
- Created a system that feels helpful without overstepping boundaries
Most importantly, we built something that reflects how college actually works.
What We Learned
We learned that AI becomes powerful only when it’s grounded in context. Intelligence without understanding leads to noise. Systems that understand environment, intent, and human boundaries can do far more with less AI.
We also learned that:
- Social platforms don’t need more feeds—they need structure
- AI shouldn’t replace people—it should organize them
- Trust is a feature that must be designed, not assumed
Bonded reinforced our belief that the future of AI is context-first, system-driven, and human-centered.
What’s Next for Bonded
Next, we plan to:
- Expand Link’s ability to guide students through transitions (first-year, major changes, internships)
- Deepen campus integrations with advisors and organizations
- Improve onboarding and schedule parsing
- Continue refining Linkya as a reusable intelligence architecture
Bonded is just the beginning.
College already has the people, the data, and the opportunities.
It just never had a system that understood them.
Bonded doesn’t just show college.
It helps you navigate it.
Built With
- expo.io
- fastapi
- llmindex
- python
- railway
- react-native
- supabase
Log in or sign up for Devpost to join the conversation.