Tools using artificial intelligence (AI) are helping fixed income traders to get better execution and to support investors. While ChatGPT kicked off the conversation on the high street, innovations like Bond GPT, launched by LTX two years ago, have brought it to Wall Street.
Isabel Carman and Katie Savignano, directors for relationship management at LTX, tell Trader TV how their clients have moved from AI-curious to AI-active through positive engagement with AI tools. They outline the the expansion of these services into areas such as Bond GPT Intelligence to support different functions and stakeholders, and the increased capacity to integrate AI functionality into the trading workflow.
Transcript:
Jo Gallagher: Welcome to Trader TV, your insight into institutional trading. I’m Jo Gallaher and I’m joined by Katie Savignano and Isabel Karman at LTX to discuss the growing functionality of AI tools on the fixed income desk and where they’ve been applied to real world use cases. Hi Jo, it’s great to be here. Thanks for having us. So to start us off, what is the appetite for AI tools on the buy side desk.
Katie Savignano: AI has been on everyone’s mind since launch of ChatGPT. And for us, especially since launching Bond GPT two years ago, we’ve been hearing a lot of questions from clients around how to approach AI and trading. A big theme is the buy versus build conversation. We’ve seen clients use scarce resources to build entire AI task forces to develop groups of concept test EPIs and look for ways to integrate AI into their stack. But firms are realizing how complex it is to build something from scratch, especially aggregating the right data and putting it cleanly into existing workflows.
Isabel Carman In our conversations with clients, AI has shifted from a curiosity to a serious priority, but the challenge is separating hype from real utility. And honestly, this is why the response to Bond GPT has been so strong. Clients can see right away that it’s not just another chat bot. It’s trained specifically on corporate bond data and built to think exactly how traders and PMs do, how they ask questions and consume information. It also already understands bond market jargon. And training conventions, which makes adoption really simple. With Bond GPT+, clients can actually bring their own data into the mix, combining our models and compliance infrastructure with their proprietary data. This gives clients a way to move quickly without sacrificing control. And instead of building from scratch, they can just plug into a secure, flexible system that can be completely tailored to their needs and connect via API or our UI.
Katie Savignano And that’s all backed by our patented multi-agent architecture where multiple AI models are collaborating and even challenging each other to produce more accurate and well-rounded answers. This all helps clients see Bon GPT as both a secure and scalable way to bring AI into their business without the risk or resource strain of building from the ground up.
Jo Gallagher Some see AI systems as black boxes, so why use AI like Bond GPT over traditional tools?
Isabel Carman Yeah, well skepticism is totally fair and honestly it’s healthy. AI is evolving very quickly so firms are right to ask tough questions. That’s exactly why we built Bond GPT with responsible AI principles from day one. We put in strong guardrails around data security, compliance, and accuracy which as a broader company are non-negotiables. And client data is never used to train any of our models or outside of its intended workflow. Unlike some AI tools, Bon GPT is not a black box. So it actually explains to users how it got to each answer through our step-by-step patented show your work feature. We’ve also reduced the AI’s ability to think generatively which prevents hallucinations and keeps results grounded in real data. Lastly, because it’s built on broad range infrastructure, Client’s data is protected at every step. It’s this combination that’s really driving adoption. And let’s not forget, there was a time when people were hesitant to even use calculators on the trading floor. Now, of course, we’ve gone way past that. We think AI is going in the same direction towards broader acceptance across trading. So firms that lean in early and responsibly are going to have a major edge.
Jo Gallagher So how can AI be effectively integrated?
Katie Savignano Integrated into…
Jo Gallagher Workflows.
Katie Savignano One age-old problem is clients can’t find the extra screen space or spend even five minutes on another application. So we built Bond GPT Intelligence. Now Bond GPT Intelligence is about meeting traders where they are. Instead of asking them to learn a new tool or jump into a separate system, we’re embedding AI directly into the workflows they already rely on, like their watch list, trade water, and trade ticket. What makes this different than existing AI workflows is how proactive this is. As a trader navigates the platform, Bond GPT Intelligence is anticipating what they might need next, whether that’s bit-ass context, recent trade data, similar bonds, or insights from Trace. It brings forth that information in real time right when it’s most relevant in the form of seamless and platform prompts. So instead of disrupting the workflow, we’re enhancing it. Traders don’t have to go back and forth between systems so they can make faster, more competent decisions. Which is a huge step forward with integrating AI into the daily trading process.
Jo Gallagher So what’s next? How do you intend to build up Bond GPT’s functionality further?
Isabel Carman Well, like Katie said, we’re continuing to expand Bond GPT’s reach across the platform, we are layering in new data sets, refining the model’s responses and further deepening the integration into the trading workflow. We’re also hyper-focused on client feedback and learning how traders are actually using the system. So what’s working, what’s not, and how we can better support them. It’s this kind of input that’s really going to keep Bond GPTs ahead of the curve, not just adapting to the market, but really shaping where it’s going.
Katie Savignano And we’re seeing a ton of momentum around Bond GPT Plus, the version of Bond GP T where clients bring proprietary data into the experience. And we’re expanding into new fixed income asset classes. This is all with the goal of making AI an embedded part of the entire investment lifecycle.
Jo Gallagher I’d like to thank Katie and Isabelle for their insight, and of course you for watching. To catch our other shows including Traitor TV this week, every Monday at 6.45am UK time, go to TraitorTV.net.
