Coach Node - Hackathon Submission
Inspiration
After building AI-powered apps, I kept hitting the same problem: how do you let experts share their coaching methodology at scale while protecting user privacy? Every AI coach I tried forced users to choose between personalization and privacy. I wanted coaches to be as shareable as courses, but with AI personalization - without the creator ever seeing user data.
What it does
Coach Node is a marketplace for AI coaches. Experts create coaches trained on their frameworks and knowledge. Users discover these coaches and get personalized guidance. The key innovation: we separate shared coaching methodology from private user conversations. When you install someone's coach, you get their expertise - they never see your data.
Real examples: A productivity coach packages their GTD framework. A fitness trainer builds a marathon training coach. A startup founder shares their decision-making process. Users get personalized help, creators never access the conversations.
How we built it
iOS app with SwiftUI, Python backend with FastAPI, Agno framework for AI agent management. The memory system uses Qdrant vector database for long-term memory storage and retrieval - coaches remember past conversations and continuously learn from interactions while maintaining the separation between shared coaching knowledge and private user data.
Challenges we ran into
Privacy vs personalization: Coaches need context to help, but users need privacy guarantees. Solved by architecturally separating what gets shared (methodology) from what stays private (conversations, progress, goals).
Long-term memory that scales: Making coaches remember hundreds of conversations without losing context or exploding costs. Implemented semantic search with Qdrant to retrieve only relevant past interactions.
Building a dual marketplace in 3 weeks: Creating experiences for both coach creators and end users simultaneously, each with different needs and workflows.
Accomplishments that we're proud of
- Solved privacy without sacrificing quality - users get personalized coaching, creators monetize expertise, nobody compromises
- Built a functioning two-sided marketplace in the hackathon timeframe
- Made expert coaching accessible 24/7 at a fraction of traditional costs
- Coaches that actually remember and improve over time through long-term memory
What we learned
Privacy can be a product feature, not just compliance. By separating shared knowledge from private conversations, we made it possible for creators to share their expertise widely while users maintain control of their data. Vector databases aren't just for search - they're essential for building AI that truly learns from interactions.
What's next for Coach Node
Launch the marketplace with ratings and revenue sharing. Add habit tracking and decision journaling workflows. Build mobile-first distribution. The vision: become the App Store for AI coaching. If you have expertise worth sharing, you should be able to package it as an AI coach and reach millions while users maintain privacy.
Log in or sign up for Devpost to join the conversation.