The Finance Startup Bringing Agentic AI to Wall Street

ModerML is an AI workspace designed for financial services. We built a workspace similar to the Office suite, offering our own versions of Word, PowerPoint, and Excel. The key difference is that it's built on an agentic system, mirroring the resources a human has at the firms we work with. Specifically, if you're a human at FirmX, you'd have access to your files, folder systems, emails, CRM, data vendors, real-time public information, public filings, and internal custom datasets. We call this a cognitive architecture.

Today, many data providers rely on humans to extract information from public filings and convert it into a structured format. However, our observations indicate that models are already more accurate than humans for these types of tasks. While it may take some time for confidence in this shift to grow, data gathering and presentation tasks are already being fully automated at leading firms.

Last year was a year of proof of concepts. Now, our average contracts span multiple years. I believe people are increasingly looking for shorter-term or instant value. Regarding firms, including law firms, historically not buying software, I think that's true. The significant difference now is that this initiative comes from the very top; it's the number one priority on everyone's agenda.

You can definitely be calmer, but there are certain things you just cannot prepare for. No matter how much you tell yourself something is normal, there will always be unexpected issues. These could be employee-related or product-related, but they will still surprise you. The fundamental thing we realized and brought into this company is the concept of perseverance. This isn't blind perseverance, but true perseverance. There's a clear difference: if something logically makes sense, and you approach a problem unemotionally, it logically makes sense.

Customer feedback wasn't just real-time; it was collected with every order. We knew what was working and what wasn't. This meant that when we implemented drivers, we knew exactly how long it would take from the warehouse to each house because we had done the route many times before. Across all three businesses, if you establish and maintain customer-centric foundations, things tend to go well.

We aren't trying to conform to what people are accustomed to. Instead, we're considering a shift based on current observations. Previously, humans initiated processes within these systems by clicking 'run.' We believe the biggest change this year is that even this manual element will disappear, making user interface elements less crucial. In essence, these tasks will become entirely autonomous. You'll arrive in the morning, and the actions you once had to trigger will already be completed.