Analyze
Portfolio holdings are evaluated using deterministic analytics for value, return, concentration and performance.
01 / AI · GENAI · FULL-STACK
Zerodha AI
Financial Intelligence Platform
A full-stack financial intelligence platform combining deterministic portfolio analytics, controlled data access through MCP, Gemini-powered interpretation, structured validation, recommendations and audit telemetry.
PROJECT TYPEAI / Financial IntelligenceSTACKNext.js · FastAPI · PostgreSQL · MCP · GeminiFOCUSGrounded AI · Analytics · Governance
01
Problem & approach
The problem
Portfolio dashboards can show numbers without explaining what those numbers mean. The project focuses on turning portfolio and market context into structured, explainable intelligence.
The approach
Keep financial calculations deterministic, retrieve approved context through controlled tools, use the LLM for interpretation, then validate the generated response before presenting it.
02
System flow
PORTFOLIO DATA→ANALYTICS→MCP CONTEXT→GEMINI→VALIDATION→INSIGHT
Retrieve
Approved portfolio, analytics, market and news context is accessed through the MCP layer.
Interpret
Gemini turns structured context into readable portfolio analysis rather than performing the underlying financial calculations.
Validate
Schema, grounding and policy checks help keep generated analysis structured and traceable.
03
Product screens
04
Engineering highlights
DETERMINISTIC ANALYTICS
Portfolio metrics and recommendation logic remain controlled by application logic.
GROUNDED AI
The AI workflow uses structured application context instead of relying on an unconstrained response.
GOVERNANCE
Validation, compliance views and execution records make the workflow easier to inspect.
ROLE-BASED UX
Separate user and administrator experiences expose the relevant intelligence and operational controls.