Adam De Rose at SS&C Technologies (now at Boltzbit) says fixed income trading is undergoing a major technological transformation, driven by advances in data integration, execution management systems (EMS), and the rise of agentic AI. Speaking with Trader TV, De Rose shares his views on the role of EMS platforms and interoperable third-party systems in helping traders aggregate fragmented liquidity and data sources across multiple venues.
Turning to artificial intelligence, De Rose looks at advancements in Agentic AI and where he sees major opportunities emerging in being able to integrate advanced large language models into order and execution management systems, for data querying, for portfolio allocations, and supporting execution.
Interview
Josephine Gallagher – I’m joined by Adam De Rose at SS&C to discuss the evolving role of technology across fixed income from EMS to agentic. Ai
Adam De Rose – EMSs have been around for a long time right on the equities, futures, options side, and they’ve really been the backbone of a trader’s workflow. When it comes to fixed income, it’s the same story, right? It’s to try and bring fixed income into that world. In particular, it’s to do with things like aggregating liquidity, aggregating connectivity, and to try and help make sense of the chaotic puzzle that we have at the moment of loads of different venues, loads of different protocols and loads of different liquidity sources. There is a massive explosion number of data products and data vendors out there. The first challenge is trying to figure out which ones are we going to back essentially. Especially when, like us, you’ve got over 250 clients trading fixed income, and that spans hedge funds to asset managers to wealth managers. The next thing we have to solve is a lot of these products are expensive, both for us and the end client, and they’re typically gated for the very largest clients. So actually being able to move forward with that is challenging. Then once you do that, you’ve got to consume that data and display it in a meaningful way. One of the ways that we’ve addressed that is through what we call solutions marketplace, which is essentially like an app store. So inside the OMS, and now more recently, inside the EMS, we allow third party vendors to drop their widgets within our applications, and it has contextual awareness through something called the FTC3 protocol, which basically passes symbols and positions around between the different applications. So that allows the vendor providing that to display the information in a way that they think it makes sense to the end user. When we talk about actually integrating the data itself into our application, that’s when it gets a lot harder, because even though everyone now has APIs, everyone’s using different taxonomy, different delivery mechanisms, and so every integration is going to be bespoke. So we spend a lot of time trying to figure out what is the ROI on this different data platform and how many different clients are interested in it, and really trying to find the one that’s the most urgent and most pervasive.
Josephine Gallagher – Agentic AI is a big topic right now. Where do you see the greatest opportunities for its use?
Adam De Rose – Agentic AI is essentially the ability for a large language model to have agency. We as human beings have agency, which is the ability to make and affect decisions. And until recently, large language models didn’t have that capability. You could ask them questions and they could answer it, but they couldn’t take any action. Now they can call tools, and there’s something called the MCP protocol, which stands for Model Context Protocol, that was developed by Anthropic out in November. And since then, the world of MCP tools has exploded. And what we can now do is take things like our OMS APIs and turn that into an MCP server. So I’ll give you an example. I can speak to my large language model hooked up to my OMS, or potentially my EMS, via an MCP server, and ask it, can you get all my portfolios and show me where I have excess cash, and it’s going to go and retrieve that data. Okay, great. I can see that. Now I want to place this trade, and I need to allocate it in a method that I’ve never used before. Can you come up with some suggestions for that? And it will actually create the allocation, and then I can ask it to inject that back into the OMS or EMS. That’s one example. What I’m very excited about is building sort of conversational interactivity with our products, but there’s loads of other ways that we’re using it as well. For example, we’ve created an environment called Tarvos running on SS&C clouds, which our own private data center. And we’ve got the Llama model from meta, which is open source model, and that’s running inside our own private GPUs. We have guardrails, we’ve got prompt injection prevention have got hollowed governance around it, and we’re making that platform available internally to developers to build solutions for internal problems, but also external problems. So on the internal side, things like reading credit agreements and digitizing those into actual securities. And on the external side, it can be a reconciliation agent. It could be an MCP tool to integrate into the OMS, that kind of thing. So the company is very focused on AI, but delivering secure AI in a controled play.
Josephine Gallagher – I’d like to thank Adam for his insight, and, of course, you for watching to catch your other shows, including trader TV this week, every Monday at 6:45am UK time go to TraderTV.net.
Catch Trader TV shows each week:
To watch Trader TV This Week shows go to – Video
Or make sure to follow us on LinkedIn – Trader TV’s LinkedIn
Available as a podcast – Link
