The Client BRDGE Insights LLC is building Strēm — a context-aware AI investment research platform for analysts, portfolio managers, and sector researchers. The company was founded by Torrence Jennette, an investment professional, and Mallory Musante, a marketing analytics expert. Together they identified a problem most people in finance quietly accept as reality: the research process is fragmented. The Business Goal BRDGE Insights wanted to build Strēm. A platform where analysts could ask complex research questions in plain English and get answers backed by verified data, without pulling it together manually. Their goals going into discovery to see if it’s possible to: Reduce routine analysis time by 40%, freeing analysts for higher-value work Automatically track and update financial data for 3,000+ US companies on a daily basis Build a natural language interface so users could ask about financial data Create a demonstrable product quickly enough to attract pilot clients and strategic partners
Discovery Phase Outcome The discovery phase confirmed the technical feasibility of the platform and produced foundation for development. We confirmed that a system of specialized AI agents, each responsible for a different type of analysis, is the right approach. We also mapped out everything the platform needs to do and each feature described precisely enough that the development team knew exactly what to build.
Discovery Phase: 3 Key Stages The discovery kicked off with sessions between COXIT’s BA, Tech Lead, and UX Designer and the BRDGE Insights founders. The focus wasn’t on documenting requirements, it was on understanding the actual workflow of an investment analyst and finding where automation could make a difference.
Problem Mapping and Persona Research Before designing anything, the team needed to understand who would actually use the platform and what their daily workflow looks like. Three types of users were identified: analysts who research individual companies, portfolio managers who track performance and risk across investments, and researchers who scan the broader market for opportunities. Each has a different job, different questions, and a different idea of what a useful answer looks like. Understanding this upfront determined two things: the platform needed separate AI agents for different types of analysis rather than one model trying to do everything, and each user type needed to see only the features relevant to their role. At the same time, the team assessed the financial data landscape — what sources were available, how reliable they were, and what would be needed to keep data current across 3,000+ companies every day.
Building the Product Blueprint The team translated research into a list of tasks organized by section. These covered everything from data gathering to portfolio management. We defined over 80 features and described the final result for each one. This work allowed for smart prioritization. We separated the essential features for the first release from those that can wait. This helps the product launch faster to start gathering user feedback. We also mapped how the system works together before building. This helped us find potential issues while they were still just ideas. Changing a plan is free, but fixing software after it is built is very expensive. This approach keeps the project on schedule and protects the budget.
Creating the Technical Blueprint We turned research into a technical blueprint. It shows how every part of the system works and how data moves through the platform. We documented it so both technical teams and business owners understand the plan. The core decision was to use specialized AI agents. Instead of one model doing everything, each agent has a specific job. An orchestrator directs questions while a synthesis agent combines the results into one answer. This structured plan prevents confusion during development. It ensures each part of the system is reliable and can be expanded as the platform grows.
What Discovery Made Possible The discovery phase protected the project budget by identifying complex financial and technical risks beforehand. We resolved how the system handles compliance and tracking for different types of stocks to ensure the platform is legally sound. This preparation allowed the team to build the entire platform on a solid foundation from the start. By mapping out how AI agents communicate early, we avoided major system failures that would have been very expensive to fix later. This approach keeps the project on schedule and ensures the final product is a tool that professional investors can trust.
