Financial AI-Powered Chatbot
Through my internship at Publicis Sapient, I served as the sole product manager on an intern project team with five software engineers. We were tasked with designing and building an AI-powered chatbot providing investment advice for clients of a leading financial institution. My responsibilities included competitive analysis, ownership of the product roadmap, design input, leading daily scrum meetings, and more.
Project Type
Product Management
Date
June - August 2025
Skills
Business Analysis
User Experience Design
End-to-End Product Journey
As the product manager on the project, I led the team through the entire product lifecycle using Agile methodology.
Problem Statement
​Non-expert investors struggle with digesting the complexity and volume of investment information available on [financial institution]’s website, leading to difficulty making informed decisions about investing with the company.
User Personas
In order to narrow the scope of the project, I developed user personas drawn from the clients identified in the problem statement. The two different personas reveal a variety of goals and pain points to be considered as the product is developed.
Research
​At the beginning of the project, I conducted both internal and external research into existing tools. The company already has some AI investment advice tools; however, they are primarily for employee use. Further, the client-facing tools do not incorporate generative AI. In competitive analysis, I identified that some competitors are in the process of developing client-facing AI-powered investment advice, but they do not fully exist yet. There were some features used in competitive products worth incorporating into our product, such as suggested prompts and personalization. Most importantly, this research confirmed that there is a gap in the market for an AI-powered investment advice chatbot, particularly for non-expert clients.
Our Solution
Finn the Financial Assistant is a one stop shop for non-expert investors providing them personalized investment information and access to [financial institution]'s tools. Not only is Finn a stand-alone chatbot accessible through [financial institution]'s website, but he is available to provide clarification on [financial institution]'s existing informational webpages.
Consolidation
Finn has the ability to generate responses that combine historical data, [financial institution] educational information, and users’ financial data
Speed
Finn reduce the time users spend researching across numerous sites and streamlines navigation to invest money with [financial institution]
Cost Efficiency
Finn reduce the load of financial advisors by taking on a large volume of non-experts’ common questions
Technology
Tech Stack
AI Integration
AI was the right choice for this product because it scales personalized support without requiring live advisors and improves the target demographic's user experience by translating complex financial terms into understandable human language.
​
Finn uses generative AI (LLMs) to answer financial questions, explain concepts, and engage naturally with users. Finn also uses Retrieval-Augmented Generation that pulls relevant information from a curated knowledge base to enhance responses. He is augmented with [financial institution] internal content, user data (when logged in), and page context. Prompt engineering was used to narrow Finn's scope and ensure he remains compliant, as well as fine-tuning his friendly personality.
User Testing
We ran user testing on five individuals regarding intuitive user experience in order to iterate and improve Finn. One user asked Finn questions they had asked a paid financial advisor and received similar answers. Other notable feedback from users led us to minimize the amount of features shown at once on the landing page, as well as prompt engineering to make Finn's overall responses shorter and more digestible.
Accessibility
Finn passes WAVE and aXe accessibility tests, is fully keyboard accessible, and includes "Skip to Main Content" and "Skip to Chat".