- Precision and recall trade-offs
- Investigator experience and explainability
- Governance and audit trail
- Measuring operational impact
Archives: Agenda
EU AI Act and DORA in practice: building an AI control framework that works
- Classifying use cases and defining high risk in practice
- Documentation and monitoring that survive audit
- Third party AI risk and contractual obligations
- Translating regulation into operating controls
The hidden costs of AI adoption
- Maintenance and model updates
- Governance and compliance overhead
- Talent and organisational costs
- Vendor lock-in risks
Model risk meets GenAI: validation frameworks that hold up to scrutiny
- Validation for probabilistic systems
- Robustness and bias testing
- Monitoring and re validation cadence
- Aligning traditional model governance and GenAI
From sandbox to production in 90 days: execution lessons from the front line
- Use case selection and kill criteria
- Integration with legacy systems
- Human oversight and escalation paths
- Measured impact and post launch governance
I, Robot: Governing intelligent systems
We examine how firms are translating principles into enforceable controls, and where governance may be slowing adoption or shaping competitive advantage
- Defining and operationalising AI risk appetite
- Embedding controls at design time, build time and run time
- Trade-offs between speed, value and control
- What regulators expect and where firms are over-engineering
PANEL DISCUSSION: Do Androids Dream of Electric Sheep? Synthetic data and autonomous agents
As AI begins to act on behalf of humans, we explore how customer models, distribution and control of the interface are fundamentally changing. Machine customers and synthetic users: who owns the relationship?
- Rise of agent-mediated interactions and delegated decision-making
- Whether banks retain or lose control of the customer interface
- Detecting human versus machine behaviour in financial journeys
- Implications for product design, distribution and monetisation
Foundation: A GlobalData perspective
- What “inevitable” actually means in AI adoption
- Separating signal from noise in vendor and internal claims
- Where to place bets and where to hold back
- Building institutions that can adapt, not just predict
PANEL DISCUSSION: The data bottleneck: why AI strategies fail before they start
- Data quality and availability challenges
- Governance and ownership issues
- Integration with legacy systems
- Building AI-ready data foundations
Designing services for AI agents, not humans
- API-first interaction models
- Structured vs conversational interfaces
- Authentication and identity challenges
- New service design principles