The AI Advantage: Use Cases, Governance, and Practical Innovation
What does the AI Revolution Mean for Asset Management?
As TradeTech Europe 2025 opened its doors to a packed room, the conversation turned immediately to one of the most transformative topics of our time: artificial intelligence in asset management.
Dr. Iro Tasitsiomi, a former astrophysicist turned AI leader at T. Rowe Price, delivered a compelling keynote that combined technical depth with visionary insight, followed by a candid interview with trading veteran Marc Wyatt.
Dr. Tasitsiomi traced the evolution of AI, from its post-war origins and early “AI winters” to the modern explosion of generative models, large language processing, and machine learning.
She outlined how AI is now reshaping financial services, citing figures such as $16 trillion in projected economic contribution by 2030 and a 2x increase in generative AI use by asset managers since 2023.
Her central message: this wave is different—not just technologically, but societally. AI has shifted from niche to ubiquitous, demanding industry-wide fluency.
Crucially, she emphasized that AI's real value lies between three poles: automation (eliminating mundane tasks), augmentation (informing better human decisions), and differentiation (enabling unique capabilities).
Examples included using AI for content summarization, predictive insights, and smarter client engagement, with the caution that adoption must come with clear governance, especially around data privacy and model transparency.
Marc Wyatt then offered a grounded view from the trading desk. His team is already deploying AI to curate news and data feeds for traders, accelerate internal reporting, and support execution decisions in FX via pattern recognition across vast TCA data sets.
Rather than replacing traders, AI is freeing them to focus on high-impact decisions. A phrase that resonated: “Someone using AI may take your job—not AI itself.”
Dr Iro Tasitsiomi, Head of AI & Data Science, T. Rowe Price, and Ex- Astrophysicist
The discussion also acknowledged the risks: the need for explainability, guardrails, and skepticism about “black box” tools.
Both speakers warned against overhyping the technology or delegating accountability to machines. Instead, they advocated for collaborative integration—empowering humans with AI, not supplanting them.
Their final thought: firms must be willing to redefine value. Many of AI's biggest gains come not from headline ROI but from subtle productivity improvements, smarter processes, and better resilience—benefits that demand a shift in mindset as much as in models.
AI in Trading Today: Beyond the Hype
At the heart of TradeTech Europe’s AI discourse was a compelling keynote interview with Mathias Eriksson of AP2, Sweden’s state pension fund, offering a rare look into real-world AI adoption within a lean institutional trading team.
Moderated by Virtu's Rob Boardman, the session peeled back the layers of AI implementation—focusing not on future possibilities, but on what’s working now. Eriksson, overseeing a three-person trading desk responsible for $25 billion in annual equity turnover, shared how AP2 deployed an AI-based algo wheel, developed in partnership with BTON and integrated via Virtu.
The objective: to enhance efficiency in broker selection while adhering to bundled commission requirements. Rather than choosing execution strategy, the AI recommends which broker to route an order to, based on historical trade data and existing commission targets.
While Eriksson acknowledged that strategy selection has greater economic impact, he explained that broker allocation was a pragmatic starting point. The AI tool continuously ingests past and current trades, learning iteratively and supporting smarter, data-informed decisions. Importantly, AP2 complements the AI output with manual oversight and continuous performance evaluations—including plans to randomize a portion of trades as a control group for bias detection.
The conversation didn’t shy away from concerns. Eriksson highlighted the importance of governance: secure data sharing, NDA protections, and vendor relationships grounded in mutual trust. Interestingly, the AI’s "black box" nature wasn't a deterrent—after all, as Eriksson noted, traditional algos are often just as opaque. The real value lies in consistent outcomes and time savings, enabling traders to concentrate on complex orders rather than routine flows.
TradeTech Polling: What is your biggest concern around using AI in trading?
Regulatory or compliance risk
Lack of explainability
Cost or resource required to implement
Performance inconsistency
I have no concerns
TradeTech Polling: What's the biggest enabler for safe and scalable AI in your firm?
Better data quality and infrastructure
Upskilling/staff education and training
Clear ROI / business value demonstration
Stronger governance and compliance frameworks
Executive buy-in and budget
Audience polling underscored the community’s diverse concerns - ranging from regulatory clarity and auditability to implementation costs and vendor dependence. Eriksson responded with clarity: at AP2, the biggest hurdles are internal resources, not risk aversion. Legal and compliance teams have been supportive, viewing AI as an opportunity rather than a threat.
Crucially, Eriksson framed AI not as a disruption, but as an incremental evolution. AP2’s culture, while cautious, encourages experimentation - and that ethos has helped them punch above their weight in innovation. The project has not only delivered measurable P&L benefits but also improved broker relationships by standardizing and smoothing trade flow distribution.
In closing, Eriksson offered a pragmatic vision for AI adoption: begin with manageable use cases, measure continuously, and focus on long-term relationships—with both vendors and data. AI, he asserted, isn’t replacing traders—but it’s definitely reshaping their world.
