How Theorem Thinks About AI

There is useful work for AI in financial operations: configuring workflows, investigating problems and helping make sense of unfamiliar data. But giving a model more work should not automatically give it more access or authority.

In this paper, Paul Geiger explains the thinking behind Theorem’s approach. We see an opportunity to make our software easier to use and more capable while keeping financial data, permissions and execution under application control. We also see a role for clients’ own agents, with clients choosing what authority to delegate.

Inside the paper

The Opportunity Is Real

Where AI can do more than save a few minutes, from helping set up processes to handling more of the work around them. The aim is to build on what already works, not replace it simply because AI is available.

AI Introduces Risks We Have to Address

What happens to financial data when a model is involved? What is it allowed to do, and what happens when it is wrong? These questions shape how we think about incorporating AI into post-trade workflows.

Theorem’s Direction

How we envision models working within application controls, without receiving private client financial data. The paper also explores how authorized client agents could use Theorem’s capabilities without firms having to rebuild the underlying software.

At Theorem, we are skeptical of the hype. We are not skeptical of the technology.

Download the paper

Read the full document for a closer look at the opportunities, the boundaries and what our direction means for clients.

Download the paper →

AI disclosure: AI assisted with drafting this page.