BFSI & FinTech WeWinSubmitted July 26, 2026

Proactive Financial Orchestrator for the Financially Included, Digitally Guided

An MCP app on the Model Context Protocol built by WeWin at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon and deployed on NitroStack.

About this project

Most "AI financial advisors" are a chatbot bolted onto a bank app: you ask a question, an LLM answers from whatever context it's given, and the moment you close the tab, it stops thinking about you. We Win is built to be structurally different — a system that watches, calculates, and acts on a user's behalf continuously, while the AI itself is never trusted to touch a single number. Deterministic core, not a chatbot guessing. Every score you see — the six-dimension Financial Health Score, the Safe-to-Spend limit, the Credit Trust Score — is computed by hard-coded, testable Python engines benchmarked against RBI, SEBI, and IT Act standards. The LLM layer (Google Gemini, orchestrated via LangChain and CrewAI) is architecturally forbidden from performing arithmetic; it only ever narrates numbers that have already been verified. This isn't a prompting choice — it's a hard boundary, which means the system cannot hallucinate a financial figure by construction, not by hoping the prompt holds. A crew of agents, not one model doing everything. Instead of a single LLM handling every task, We Win runs six narrowly-scoped agents — Aggregator, Context & Incident, Market Watch, Loan Matching, Explainer, and Execution — each with exactly one job and exactly the MCP tools needed to do it. None of them ever holds a database credential. Every read and every write is mediated through a purpose-built MCP tool, so a hallucinated or malformed request from an agent can never reach Supabase or Neo4j directly — a non-negotiable requirement when the data is financial. Genuinely proactive, not reactive. A static balance tells you what you have; it doesn't tell you what you can safely spend after tomorrow's EMI and next week's rent. We Win's Safe-to-Spend Guardian nets obligations against income in real time and alerts users before they overdraw — while a life event (a new baby, a job change) instantly re-weights only the affected part of their plan, without re-running the whole system. Not

BFSI & FinTech track

Build AI solutions for banking, payments, insurance, fraud detection, lending, and financial inclusion.

Team WeWin

  • Darshan RLead

  • Sanjesh Nandha K

  • Tarika M

  • Daggubati Viritha

Frequently asked questions

What does Proactive Financial Orchestrator for the Financially Included, Digitally Guided do?
Most "AI financial advisors" are a chatbot bolted onto a bank app: you ask a question, an LLM answers from whatever context it's given, and the moment you close the tab, it stops thinking about you. We Win is built to be structurally different — a system that watches, calculates, and acts on a user's behalf continuously, while the AI itself is never trusted to touch a single number. Deterministic core, not a chatbot guessing. Every score you see — the six-dimension Financial Health Score, the Safe-to-Spend limit, the Credit Trust Score — is computed by hard-coded, testable Python engines benchmarked against RBI, SEBI, and IT Act standards. The LLM layer (Google Gemini, orchestrated via LangChain and CrewAI) is architecturally forbidden from performing arithmetic; it only ever narrates numbers that have already been verified. This isn't a prompting choice — it's a hard boundary, which means the system cannot hallucinate a financial figure by construction, not by hoping the prompt holds. A crew of agents, not one model doing everything. Instead of a single LLM handling every task, We Win runs six narrowly-scoped agents — Aggregator, Context & Incident, Market Watch, Loan Matching, Explainer, and Execution — each with exactly one job and exactly the MCP tools needed to do it. None of them ever holds a database credential. Every read and every write is mediated through a purpose-built MCP tool, so a hallucinated or malformed request from an agent can never reach Supabase or Neo4j directly — a non-negotiable requirement when the data is financial. Genuinely proactive, not reactive. A static balance tells you what you have; it doesn't tell you what you can safely spend after tomorrow's EMI and next week's rent. We Win's Safe-to-Spend Guardian nets obligations against income in real time and alerts users before they overdraw — while a life event (a new baby, a job change) instantly re-weights only the affected part of their plan, without re-running the whole system. Not
Who built Proactive Financial Orchestrator for the Financially Included, Digitally Guided?
Proactive Financial Orchestrator for the Financially Included, Digitally Guided was built by team WeWin at the Amrita University Coimbatore NitroStack × MCP To The Moon hackathon, in the BFSI & FinTech track.
What is an MCP app and how is it built?
An MCP app is an application built on the Model Context Protocol — an open standard that lets AI agents connect to tools, data, and APIs. This project exposes MCP tools and resources that agentic AI systems can call. It was built and deployed on NitroStack, the full-stack platform for shipping MCP apps and servers.