Overview

LSEG Workspace launched its first generative AI feature for external pilots — a major milestone in transforming how financial professionals interact with data. NL2Wealth introduces a natural language layer that allows Wealth advisors to discover, visualise, and share insights simply by asking questions in English.

As the Product Manager for NL2Wealth, I led the strategy, and cross-functional alignment behind this new capability, shaping how natural language fits into the broader Workspace ecosystem.

Problem

Workspace offers rich financial data across multiple apps, screens, and navigation paths — but this power comes with complexity. Advisors often spend significant time locating relevant datasets, switching between tools, and piecing together insights manually.

At the same time, user expectations are shifting.
Traditional click-and-navigate workflows are being replaced by simple, conversational interactions powered by generative AI. Users now expect platforms to understand questions, capture context, and return the right insights instantly.

Natural Language to Wealth was created to bridge this gap.

Role

• Defined the end-to-end natural language experience for Wealth.

• Identified 250+ question patterns from dashboard data and created structured responses.

• Shaped intent models, context behaviors, and user flows.

• Mapped high-value advisory use cases for initial pilots.

• Ensured accuracy, safety, and compliance with Workspace standards.

• Collaborated with AI, engineering, design, and entitlement teams.

• Partnered with leadership to position NL2Wealth as a foundation for future AI experiences.

My focus: make natural language the simplest, fastest entry point to Workspace insights.

I led the product direction for NL2Wealth across strategy, UX, and cross-team integration

LSEG AI Strategy

Solution

We are building NL2Wealth as a conversational AI agent that integrates with the Orchestrator and a front-end entry point to deliver an end-to-end chat experience.

Core Capabilities of Agent

• Natural Language Understanding
  Understand plain‑English financial questions and normalize them to canonical Wealth intents.

• Intent Recognition
  Maps queries to Wealth-specific scenarios like Movers, Allocations, Events, and Daily Performance.

• Context Awareness & Switching
  Maintains and transitions context across Team → Household → Client → Account seamlessly.

• Entitlement-Scoped, Factual Retrieval
  Delivers 100% accurate, entitlement-validated data from the Wealth Dashboard.

• Tool Orchestration via MCP
  Dynamically selects and executes the right backend tools/APIs based on intent and context.

• Structured, Consistent Output
  Return clean, unified JSON aligned with Dashboard card logic

• Compliance & Security by Design
  Validates tokens, enforces PII handling, and ensures safe error handling.

• Operational Guardrails
  Optimized for speed (P95 ≤ 2.5s), reliability (<3% error rate), and strict scope (North America, “today vs previous close”).

Impact

• Improved data discoverability for advisors and validated the experience through ongoing testing with 500 pilot users - directly addressing one of Workspace’s most persistent customer needs.

• Delivered sub-xs agent response time and ~xs end-to-end latency, enabling fast, conversational access to performance, risk, movers, and book-of-business insights.

• Achieved ~xx% intent interpretation accuracy across varied natural-language queries mapping to the same underlying advisor questions, improving relevance and trust in results.

• Reduced system and retrieval errors to ~x%, ensuring reliable, entitlement-safe insight delivery across Team → Household → Client → Account views.

500

15

Years of experience

Happy clients