Conversational AI Agent for a Global Asset Management Department

Elderly couple consulting with a financial advisor

Services:

Wealth and Asset Management

Industry:

Financial Services

Location:

North America

Client since:

2016

Sombra collaborated with a top-10 global bank to create a Conversational AI agent for their Asset Management division. Designed for financial advisors, it enables plain-language queries and instant access to insights on funds, ETFs, and investment strategies — supporting faster, data-driven decisions and improved client service.

Portable across all major cloud platforms and compatible with leading AI providers: AWS, Azure, GCP, OpenAI, Meta, Antropic

Business challenge

Searching for the right fund, ETF, or financial information can be time-consuming and overwhelming. Advisors had to navigate an internal knowledge base with over 3,500 growing records —updated nearly every day — which made retrieving the right information slow and inefficient. The process was not scalable, leading to complex charts, repetitive queries, and increasing client expectations that delayed decision-making and raised support costs.

Too much data, too little clarity

Advisors struggled with information overload, leading to frustration and slow decision-making.

Time-consuming search

Locating the right fund or ETF required multiple steps, delaying client recommendations.

Complex data visualization

Interpreting charts and graphics took extra time and expertise, increasing the risk of missed insights.

Missed opportunities

Slow access to information resulted in delayed decisions and lost market chances.

High support workload

Repetitive advisor questions overwhelmed support teams, stretching internal resources and increasing response time.

Lack of scalable personalization

Tailored insights were difficult to deliver efficiently, limiting the advisor’s ability to provide personalized recommendations at scale.

AI Agent for smarter decisions

Tech stack: Azure OpenAI GPT-4o, Azure Blob, LangChain, text-embeded-3-large, PostgreSQL + PGVector, Python, GPT-4o mini

Our solution is a Conversational AI Agent that provides instant answers to natural language queries, removing the need to manually search through thousands of records. It simplifies access to complex charts, fund data, and investment insights, helping advisors respond faster and more efficiently.

Interpret visual data
The AI agent analyzes charts and graphs embedded in articles and documents stored within the internal knowledge base, extracting meaningful insights to generate accurate, context-based responses.

Reference-backed responses
Information is supported by easily traceable references, including documents and trusted sites within the internal knowledge base. This enables advisors to quickly validate information and ensure regulatory compliance.

No Hallucinations
The AI agent runs in Context Only mode by default, using only trusted internal sources to ensure accurate, verifiable responses without hallucinations. General and Auto modes can access broader data, including external sources, but may be less reliable.

Content Enriched by Metadata
We analyze feedback and chat history to categorize queries, identify trends, and uncover knowledge gaps, continually enhancing the AI agent’s performance.

Architecture

Built on Azure, but cloud-agnostic by design
We delivered a scalable, enterprise-grade AI agent using Azure technologies — designed to surface the right data, simplify complex queries, and fit seamlessly into the advisor workflow.

Business value

A business challenge turned into measurable success:

3,500+ documents supported
Powers rapid information retrieval, scalable decision-making, and reduced support costs.
30% reduction in data collection time
Boosts advisor productivity by streamlining access to critical insights.
15% increase in Client Satisfaction Score
Enables financial advisors with faster insight delivery, less manual effort, and a more trusted client experience.