Independent Investment Management Firm
Intelligently Mining Complex Content With an LLM Assistant
What if… investment strategists could use a secure GenAI interface to confidently and quickly gain insights from disparate, highly complex business resources?
Teams Needed Business-Critical Insights Faster
An independent investment management firm’s Investment Strategy Group (ISG) struggled to gather insights from multiple sources of structured and unstructured data. Years of research materials, archived as PDF files, provide an invaluable source of information but were difficult to search and siloed from the firm’s central data warehouse, which curates and harmonizes structured data from various business sources.
Combing through these different datasets was laborious. Custom BI reporting with tools like Tableau and PowerBI provided some insight; however, the reports required specialized skillsets and time to build. ISG business users needed a faster and easier way to access information for insights and answers to questions and inquiries, such as which investment funds hold a particular security or a portfolio’s exposure to a certain region in the world.
Snowflake's GenAI Tools Yield High Returns
We began by interviewing ISG users about their day-to-day operations and compiled a list of common questions that require a query of the unstructured data archived in PDF files. We compiled the same questions for the firm’s central data warehouse, which consolidates data from various systems and sources, including external partners and vendors. This discovery phase revealed information overlaps across the siloed resources.
Next, we ingested the PDF files, which were stored in various shared drives, into the firm’s enterprise data platform hosted on Snowflake.
We leveraged Snowflake’s suite of tools to power rapid data analysis and an intuitive user interface. Semantic models, Cortex Search, and Cortex Analyst work behind the scenes to mine, bridge, and contextualize the structured and unstructured data that is then optimized for a natural language response by large language models (LLMs). A conversational interface, built with Streamlit, enables ISG users to easily ask questions and receive a rapid response.
We monitored and optimized the solution to ensure the data and models yielded reliable responses to users’ various questions.
Our client’s ISG users are now able to gain greater insights and answers with a single source of accessible, trustworthy data.
Results
Snowflake Investment Pays Off in Data Dividends
Our client’s ISG users are now able to gain greater insights and answers with a single source of accessible, trustworthy data.
While extracting insights from structured data in the central data warehouse was already possible, it was manual and time-consuming. Adding the ability to search unstructured data empowers users with even more intelligence at their fingertips.
The AI-driven solution intuitively mines trusted information and generates contextually appropriate, natural language responses to accelerate team productivity and insights.
