From Automation to Execution: How AI is Reshaping Fixed Income Trading
October 22, 2026
| Trade Automation
In this Q&A, Charlie Campbell-Johnston, Head of AiEX and Workflow Solutions, and Leo Selya, Head of Algorithmic Trading, at Tradeweb, discuss how automation and AI are converging on the trading desk, and where the industry is moving from theory to practical, everyday application.
Fixed income execution has spent the last decade becoming steadily more electronic. What’s changed more recently is the speed at which automation and AI now converge into a single toolkit for traders.
Q: Automation in fixed income is not new. What’s actually changing?
Leo Selya:Automation was long framed as an efficiency play, freeing traders to focus on harder tickets. That’s still true, but it undersells what’s happening now. Done well, automation also builds the clean, structured data that powers real analysis: data that helps traders decide who to trade with, when, and through which protocol. Those decisions matter most on large, complex trades, where the cost of getting it wrong is highest. The efficiency benefit is real, but the analytical benefit is where the next leg of value sits.
Q: How is that playing out across Tradeweb’s client base?
Charlie Campbell-Johnston:We’re seeing consistent themes regardless of client type or asset class: visibility into process; confidence that each order is matched with the appropriate protocol and counterparty mix for that moment; and execution spanning multiple asset classes. Since launching AiEX in 2012, our client footprint has grown from 175 in 2020 to 343 in 2025, and average daily AiEX trades have grown more than fivefold over the same period. Automated trades made up more than 45% of institutional trades on Tradeweb in Q2 2026, up from around 30% five years ago, with adoption broadening into historically manual markets, from block trading in credit to bilateral swaps and swaptions. We’re building around those themes with tools that bring greater transparency to the rules governing automated execution, as well as smart dealer selection tools which use proprietary data to help clients with counterparty selection and price prediction.
Q: Automation adoption doesn’t happen at the same pace everywhere. What does that curve look like?
Charlie Campbell-Johnston:Adoption is evolutionary, and the starting point often differs by product. Markets with deep liquidity, standardised instruments, transparent price discovery and an established electronic franchise are naturally more predisposed to automation. By way of example, in European ETFs, automation had accounted for 94% of workflow by 2025. Similarly, automation in European government bonds rose from 39% in 2020 to 65% in 2025showing how quickly adoption can deepen once electronic liquidity and workflow coverage reach critical mass. U.S. markets are earlier on the curve having grown from 18% to 30% over the same five years. That’s real progress, and it points to significant runway still ahead as adoption matures. Lessons we learn from one asset class can often help direct us to opportunities in adjacent products.
Q: How is TARA helping clients use data and AI to inform their trading decisions?
Charlie Campbell-Johnston:TARA, the Tradeweb AI Research Assistant, is an AI-powered chatbot that helps clients access and interpret relevant data through a conversational interface. It’s increasingly the layer that helps clients understand how AI can assist with a decision before it’s made, surfacing insight, analysis and an audit trail alongside the trade. Clients are using that to build their own rules and logic, rather than accepting a fixed workflow, a bridge to automation that helps shape a decision, not just execute it.
Q: Is there a risk that automation and AI reduce the trader’s role rather than enhancing it?
Leo Selya:A fair question, one I’d expect from experienced desks. From 15-plus years on the buy-side building execution strategies, my view is that automation and intelligent execution have consistently improved trader performance rather than replaced traders. You still need someone overseeing the system, but now they’re equipped with better tools and more time to use them, instead of being buried in operational workflow. The offering also has to be modular: some clients want the full stack, others just transaction cost analytics or a clearer version of automation they already use. Collaborative innovation, solving specific pain points rather than shipping one-size-fits-all, is central to how we’re building this.
Q: Looking ahead, what’s the next stage of this evolution?
Leo Selya:There’s still real headroom. Only around 30% of the global swaps market trades electronically today, even after roughly 14% average annual growth in swaps volumes over the past decade, so the runway for automation and AI to do more, not less, is substantial. A year or two ago, this conversation was largely theoretical. Now the focus is on application: real client trends, live projects underway, and how a genuinely converged toolkit performs across asset classes rather than a single niche. That’s where the value compounds: not in any one tool, but in how the pieces work together for each client’s needs.
