Funds Congress Highlights

• Moderator: Lilia Christofi, EMEA FS Data, AI and Tech Lead, PwC
• Megan Caywood Cooper, Fintech Industry Expert
• Marcus Vinicius Martinez, Industry Advisor, Worldwide Financial Services, Microsoft
• Siobhán Noble, Chief Data and AI Officer, Carne Group
• Gerald Rehn, Head of EMEA Distribution, BNY Investments

This panel explored the leadership imperative, operating model redesign, and practical execution for AI and agentic AI transformation in financial services, emphasising the shift from experimentation to profit and loss impact and responsible deployment.

Leadership and operating model redesign: CEO-level ownership is essential, as AI transformation is a business transformation that cannot be delegated solely to technology officers. Firms must move beyond pilots and proof-of-concept graveyards to embedded, scaled impact in core workflows, with less than 20 per cent of firms currently achieving this. Operating models must be redesigned around AI and agents rather than bolting AI onto legacy sequential processes.

An agent functions as an orchestrator reacting to triggers including voice, email, or applications, equipped with authorised knowledge sources and ability to use tools. Emergent intelligence increases when agents converse, with future capabilities potentially including self-healing, requiring guardrails and human-in-the-loop oversight at defined checkpoints.

Agentic AI defined: Firms should target high-volume, high-friction, high-pattern decision areas heavy on unstructured data with clear economic return potential, using decision matrices to select appropriate tools rather than labelling everything as AI. Governance frameworks must define which decisions to delegate to AI, who is accountable when something goes wrong, and how oversight will operate, recognising that agent failures may be less frequent but systemic in nature.

Use case selection and governance: Data standards, explainability, and auditability are critical, with regulation serving as an accelerator when used as a clear operating framework. Responsible AI toolkits provide embedded explainability, fairness, and bias dashboards, with AI capable of auto-generating compliance test cases from regulatory artefacts.

Data standards and regulatory frameworks: High-ROI initial use cases include document extraction, legal drafting, prospectus processing, daily liquidity and compliance reports, end-to-end digitised client onboarding, and CRM intelligence for client coverage, delivering a triple win of reduced costs, increased revenue, and better client outcomes. One major institution demonstrated scale with safe enterprise access integrating multiple models, approximately 150 agents with assigned duties, performance reviews, and employee IDs, all monitored by humans and embedded in workflows.

Leadership imperatives include setting and owning the AI vision with bravery and pragmatism, rebalancing budgets from running the bank to changing the bank, assigning an executive owner accountable for profit and loss impact, starting with small high-friction decisions, creating safe environments for experimentation, and upskilling employees as people remain central to success.