Overview

To address inefficiencies in investment model discovery, I led the development of the AI Financial Model Recommender, an intelligent assistant that suggests the top 3 most suitable financial models for an advisor based on client characteristics, risk tolerance, performance preferences, and account type. The tool empowers advisors to deliver higher-quality recommendations faster and with greater confidence.

Challenge

LPL advisors lacked a tailored, intelligent way to search and compare financial models. Instead, they defaulted to:

• Manually browsing static lists
• Reusing familiar models regardless of fit
• Missing opportunities to align models with evolving client goals or market shifts

This created:

• A fragmented and inefficient workflow
• Suboptimal portfolio outcomes for clients
• Low discoverability of high-performing or niche models across LPL’s catalog

Solution

I spearheaded the design and rollout of the AI Financial Model Recommender, leveraging LPL’s internal model data and advisor/client attributes to dynamically surface best-fit investment models for any client profile.

Key Features:

• Top 3 Ranked Recommendations: Based on client fit, risk level, account type, and advisor history
• Advisor Input Matching: Accepts natural language prompts or structured filters (e.g., “show conservative models for inherited IRA accounts”)
• Performance & Suitability Filters: Incorporates historical performance, volatility, fees, and thematic preferences (e.g., ESG, tax efficiency)
• Explanation Layer: Justifies each recommendation with plain-language rationale (e.g., “this model outperformed its peers in down markets and aligns with your client’s retirement timeline”)
• Model Comparison Tool: Advisors can compare side-by-side key stats, trade rationale, and suitability for account types

Strategic Impact

• Shifted model selection from guesswork to guided intelligence
• Democratized access to underutilized but high-potential models
• Strengthened advisor-client trust by enabling more data-informed conversations
• Enabled product teams to observe which models are rising in demand and why

Results

• 50% faster time-to-model selection in pilot advisor workflows
• 3x increase in diversity of models selected (reduced overreliance on legacy models)
• Positive advisor sentiment in usability testing around confidence, speed, and value-add
• Recommendation engine is now being considered for deeper integration into advisor platforms and CRM workflows