Project Ekko

Inspiration

Our team noticed a recurring problem across different professional environments: the "Networking Cliff." Whether at a business case competition or a technical lab, high-value "action items" from great conversations usually vanish into forgotten transcripts or messy notebooks. We built Nodalli Echo to bridge this gap, ensuring that every casual verbal commitment actually turns into a professional milestone.

What it does

Nodalli Echo is a gamified CRM that uses AI to transform conversation into action. It ingests raw, informal speech transcripts, filters out the "fluff," and extracts structured "Quests." By completing these tasks, users earn Reputation Score (rs), turning the long-term maintenance of a professional network into a rewarding, gamified experience with real-time health tracking.

How we built it

The system is powered by a Secure Proxy Architecture. We deployed n8n self-hosted on an Ubuntu Server to act as our "middle-man" for AI processing and Airtable integration. The frontend is a modern dashboard built with Tailwind CSS, communicating with our backend via ngrok tunnels. This design ensures that all sensitive API keys remain 100% private and are never exposed to the client-side code.

Challenges we ran into

Our biggest hurdles were CORS management and data normalization. We initially struggled with the dashboard retrieving database metadata instead of actual task records. We had to pivot our n8n logic from "Base" resources to "Record" resources and develop a robust JavaScript "Universal Extractor" to handle varying JSON structures without breaking the UI.

Accomplishments that we're proud of

We are particularly proud of our Security Layer. While many hackathon projects hardcode API keys, we successfully implemented a Backend-for-Frontend (BFF) pattern. Additionally, our team successfully applied this logic during a recent business case competition, proving the platform's viability in professional networking scenarios. We also integrated a weighted mathematical scoring system where task difficulty directly influences the growth of a "Network Health" saturation curve:

$$H\% = \min\left(100, \frac{RS_{current}}{RS_{target}} \times 100\right)$$

What we learned

This project was a deep dive into Full-Stack Security. Our team learned how to manage cross-origin requests between cloud-hosted frontends and home hosted servers, handle "pre-flight" browser checks, and the importance of Static Domains for stable webhooks. We also sharpened our skills in engineering AI prompts for strict, structured JSON extraction.

What's next for Project Ekko

Moving forward, we plan to implement Native Audio Integration for direct recording within the app. We also aim to develop Automated Follow up Drafting, where the AI not only generates the task but also drafts a context-aware "Great meeting you" email to be sent immediately after a conversation is processed.