One of our IT clients, a leading financial investment firm, didn’t know we did strategic business consulting. Every week, they were spending 80 hours of analyst time working on a report that would take almost 2 weeks to be ready to deliver to clients. The information was incredibly detailed, had a wide variety of data sources (both public and internal), and included a great deal of analysis done by very smart people.
A few months ago, one of our engineers heard about this work. She immediately got one of our AI guys involved. Once the analysts (and partners) were convinced that what we were doing was secure and safe (the only way we work with AI), we came up with an AI-based tool to reduce the analyst time from 80 hours to 16, but even more important, to cut to the delivery time from 2 weeks to 4 days. The tools involved around 75 prompts and a good amount of Python and other types of code, use of APIs and other methods to both gather data and keep it safe, and feedback loops and use of multiple AI engines to detect hallucinations and AI mistakes.
The analysts still do the most intelligent parts, but so much of the time-consuming work has been eliminated. Likewise, we are now working on tools for them that looks for anomolies and patterns in data. The tools themselves are not smart, but what we are working on for them has already helped them find interesting threads to investigate. Where the first tool has saved valuable time enabling smart people to do more, the tool we are currently working on for them has already uncovered things that are proving valuable and allowing them to gain a competitive advantage in a very competitive space.
