UPL Ltd · Bangalore · July 2025 – Present · Status: ✅ Shipped
📌 TLDR
Led product strategy for a production-grade multi-agent AI platform on LangChain/LangGraph that consolidates EXIM, FX, and market signals into a real-time decision layer for Global Heads, Country Heads, and senior functional leadership — scaling to 1,000+ monthly active users month-over-month from an initial 200-person leadership base.
Company: UPL Ltd — global agrochemical company operating across 130+ countries Pod: 4-person pod (3 data scientists, 1 data engineers) My role: Product Manager Timeline: July 2025 – Present
CXOs, Global Heads, Country Heads, and senior functional leadership were making strategic decisions across multiple fragmented data systems — EXIM data, FX signals, commodity market intelligence, regional sales data — with no unified view, no real-time access, and no way to correlate signals across functions.
The consequences:
No single source of truth — decisions made on stale, siloed reports
Correlating EXIM, FX, and market signals required manual effort across multiple analyst teams
Leadership access was desk-bound — no mobile or voice access to intelligence on the move
The time between a market signal appearing and leadership acting on it was measured in days, not minutes
Strategy & vision: Defined the product vision, roadmap, and success metrics for the platform
Technical architecture: Partnered with engineering to architect the agent orchestration layer on LangChain and LangGraph, defining API contracts for real-time data integration
Rapid prototyping: Built end-to-end wireframes and functional prototypes using n8n and Lovable, translating research into user stories and engineering-ready requirements ahead of engineering build-out
Multi-modal expansion: Led the extension of the platform with Speech-to-Text and Text-to-Speech capabilities, and shipped a mobile-rendered version of the web application
Testing & validation: Designed and ran structured A/B tests across 4 agent response formats to validate what leadership found most decision-useful
Adoption: Owned the scaling of adoption from the initial 200-person leadership base to Global Heads, Country Heads, and cross-functional leadership globally