June 16th 2025; AI Agents, Trading Tech & Regulation: The Next Wave
Agentic AI lands in capital markets — reshaping trading infrastructure, accelerating agent-led investment models, transforming operational onboarding, and a global regulatory rethink — as the FCA’s Supercharged Sandbox, EU AI Act, and sovereign AI buildouts converge to redefine market structure.

1. What I Learnt from the UK’s AI Summit
   The Agents have landed. Agentic AI is advancing rapidly, turning large language models from passive chatbots into orchestration agents capable of autonomously coordinating complex, multi-step business processes. Google’s live demo, led by Daniël Rood, showcased agents chaining tasks, managing subtasks, making decisions, and integrating data across platforms. IBM calls this the rise of “Systems of Intelligence,” with 85% of its clients already experimenting and over half expecting enterprise-wide deployment within two years. Early adoption is growing across capital markets — from client onboarding, KYC, cross-border compliance, and financial crime monitoring to portfolio aggregation and investment support. But governance challenges remain: cascading errors, opaque decision chains, and audit complexity require new controls. Traditional QA is no longer enough; firms must adopt real-time monitoring, adaptive risk guardrails, explainability frameworks, embedded oversight, and evolving ethical governance. For more, see (https://www.linkedin.com/pulse/whacky-race-agentic-ai-what-i-learned-uk-aisummit-2025-rebecca-healey-pye3e) which sets out why sustainability, transparency, and democratic accountability are now central to responsible AI deployment.
2. What Agents Could Mean for Trading Tech
   AI-native, screenless devices could move traders and PMs beyond multi-screen setups into a world of real-time, voice-driven AI agents that continuously monitor markets, surface actionable signals, advise on risk and execution, and dynamically adjust to conditions. Sam Altman’s acquisition of Jony Ive’s AI hardware company (The Verge, 2025) signals an inflection point for trading infrastructure. Agents are already gaining traction across the investment process itself: Sydney’s Minotaur Capital, with its home-grown Taurient system, scans 35,000 articles weekly to deliver 23.5% YTD returns (The Australian); quant-driven VC platforms like China’s Baiont and QuantumLight in the US are leveraging “Moneyball-style” datasets for venture investing (Financial Times); while retail platforms like Robinhood Cortex and Public Alpha are synthesising news, earnings calls and social media into actionable insights (MarketWatch). 
3. FCA’s Supercharged Sandbox – What Does It Mean?
   The FCA’s launch of its new Supercharged Sandbox marks a significant move to support early-stage AI prototypes and proofs of concept. Open for applications until 11 August 2025, firms can submit proposals detailing their scope, technical needs, and innovation potential. Successful applicants will gain access starting October 2025 to GPU-powered virtual environments, synthetic anonymised datasets, Nvidia AI tools, and direct mentorship from FCA and industry experts, following a bootcamp in September. The sandbox lowers barriers to AI experimentation while embedding regulatory collaboration on consumer protection, bias mitigation, explainability, and market integrity risks, laying the groundwork for the FCA’s future AI Live Testing regime. More broadly, it reinforces the UK’s leadership in fintech, using outcomes-based regulation like Consumer Duty and the Senior Managers & Certification Regime (SM&CR) rather than rigid, bespoke AI rules. 
4. More on AI Regs from Lord Holmes
   Also at last week’s UK AI Summit, Lord Holmes and Dr. Laura Gilbert CBE (former head of 10DS and i.AI) addressed the regulatory challenge of fast-moving AI adoption in financial services. Both warned that prescriptive rules — like the EU AI Act — risk falling behind as AI becomes embedded across trading, compliance, and risk management. They advocated for a principles-based framework supported by an AI Authority: a cross-sector statutory body offering expert, adaptive oversight. Dr. Gilbert flagged the difficulty of cross-government coordination — especially relevant for financial institutions operating across jurisdictions. Encouragingly, UK policy momentum is shifting toward faster action, with growing emphasis on transparency, open-source development, and agile regulation — critical as AI agents increasingly drive trading, real-time risk, onboarding, and compliance. The urgency is underlined in the AI Now Institute 2025 Landscape Report, which warns that AI is rapidly shifting from passive tool to decision-making agent, raising acute questions of accountability, auditability, and governance in mission-critical systems. Democratic oversight and robust governance will be essential to ensure AI serves the public interest, not acts upon it. 
5. And The Big Switch Continues
   According to Morningstar, EU-based AI and big data funds have now reached $22.7bn AUM, representing nearly two-thirds of the global $38.1bn total. But while holdings remain heavily US-centric (Nvidia, Microsoft, Alphabet), an interesting switch in Europe is emerging away from US service providers to protect digital sovereignty. In Germany, Schleswig-Holstein will fully replace Microsoft tools—including Word, Teams, and Outlook—for 60,000 public servants and teachers. In Denmark, the Ministry of Digital Affairs, along with Copenhagen and Aarhus, is migrating staff from Windows and Office 365 to Linux and LibreOffice, citing geopolitical risk, with full migration expected by November. Sweden’s public sector has been advised to phase out U.S. cloud providers to avoid GDPR-related legal uncertainty (Euractiv). In France, schools and ministries are exploring alternatives under the national Cloud au Centre policy. As entire European governments exit Big Tech platforms for compliance reasons, expecting individual schools or teachers to manage such risks is increasingly unrealistic. In Europe’s public sector, the new rule is simple: no compliance, no entry.

Noting that London’s AI Summit wasn’t the only major event last week — later this week I’ll be in Paris speaking at the ACI World Congress 2025 alongside @Charles le Halle, discussing Europe’s financial market trends, infrastructure challenges, and emerging digital technologies. I’ll also be catching up on key takeaways from VivaTech 2025, where Nvidia’s Jensen Huang spoke extensively on Europe’s growing sovereign AI ambitions.

As always, thank you for reading! Let me know what you found helpful, what you disagreed with — and what you’d like to see more of.
Many thanks,
Rebecca