SignalNex

What's next for SignalNex

About the Project

What Inspired Us

The modern insurance and financial advisor is overwhelmed. They manage dozens of daily signals missed premiums, policy renewals, life events, partner opportunities, and compliance deadlines. Trying To Make Their Life Easier By Building AI Assist Them In Doing The Stuff That Usually Need Hours Of Reading Or Spending But this data is scattered across CRMs, spreadsheets, and personal memory. We realised that advisors don't need another place to store data; they need an intelligent command centre that tells them exactly what to do next. That was the birth of SignalNex.

How We Built It

We structured the app around a dual-engine architecture to balance speed and intelligence:

Frontend: React 18 and Vite for a highly responsive, component-based UI.
Deterministic Engine: A custom JavaScript rule-engine that handles instantaneous priority scoring and compliance flagging.
AI Layer: OpenAI GPT-4o-mini is used strictly for unstructured data processing—specifically, detecting knowledge gaps from advisor notes and recommending adaptive CPD paths.
Backend & Integration: Supabase (PostgreSQL, Auth) handles data persistence and Row-Level Security. We used Supabase Edge Functions to connect the app's action composer directly to the Telegram Bot API for governed client messaging.

The Challenges We Faced

Our biggest challenge was preventing the AI from "hallucinating" financial advice. To solve this, we decoupled the reasoning. The AI is only used to extract context (e.g., gap detection), while the actual Next-Best Actions are generated by our deterministic engine.

Another One Is Our Idea At The Start Didn't Really Approved From The Mentor , Since Our Idea Can Say Is Heavily AI , But It Usually Meant That AI Produce Idea usually is already in the market . So We Kinda Changed Again In Last Second For Our Idea .

What We Learned

We learned how to effectively merge deterministic business logic with non-deterministic LLMs. We discovered that AI is best used not as a all in one stuff , but as a specialised micro-service for contextual tasks like doing smaller tasks . We also gained deep practical experience configuring Supabase Edge Functions to bridge secure web apps with third-party messaging APIs like Telegram.