CDO BFSI Agenda Day 2
Receive pre-printed GenAI prompts and share tweaks and improvements with other delegates over breakfast.
Data democratisation fails when frontline staff lack literacy or breach controls, while specialist data teams drown under ad-hoc requests. A scalable talent model has become board-level priority and the only solution. CDOs must now develop controllable and accountable IT literacy to drive productivity, free up resources, and enable innovation from all business functions.
Agentic AI is the next leap beyond GenAI: autonomous digital “workers” that execute complex processes at software speed and near-zero marginal cost. Yet no UK bank or insurer has moved past pilot stage. Knowledge gaps, shaky data foundations and governance anxiety stall progress while early movers eye dramatic operating leverage. CDOs need to lay out and execute pragmatic agentic AI development and deployment roadmap before rivals out-automate them.
Legacy cores often underpin your most critical data stores, but cobbling together modern analytics on a fractured, unsupported stack kills velocity and trust. CDOs need a clear framework to decide what to retire, test safely and reinvest savings into modern data products.
Success depends on employees fluent in both business context and technical delivery – balance sheets and Python. Yet the talent market offers few such hybrid profiles. CDOs are now looking to offer credential paths, internal gig-marketplaces and reward opportunities that retain such staff without Silicon-Valley-esque pay checks.
Agentic AI will add armies of synthetic workers, yet most executives and staff still think “digital colleague” means RPA. Ignorance breeds fear and stalls investment. Structured education and change management sprints are urgent musts before scale pilots can land - but fear needs to be addressed first, and the value proposition of AI agents further explained.
Banks admit that when upstream changes corrupt millions of transactions “it’s already too late.” True lineage across multiple hops is not exactly simple, and reconciliation consumes days of analyst time while risk escalates. CDOs need transparent traceability plus fast impact assessment to cut firefights and fines.
Modernisation grinds when every new SaaS vendor faces months of privacy, security and contractual reviews – often up to 18 months. Some organisations prefer to keep a “tight ship” of pre-vetted partners because laws and global contracts make onboarding painful. CDOs must streamline vetting to keep build velocity while satisfying CISOs, CFOs and regulators.
EU, Saudi, APAC and US rules all demand local data anchors; data residency is often viewed as a competitive power play wrapped as regulation, which now shapes every architecture choice. CDOs juggling multi-cloud must prove which data lives where, negotiate exit clauses yet still keep analytics latency low. Delays invite fines - and business users waiting on reports.
Every BFSI organisation waves a “responsible-AI” memo on their intranet, yet few can prove that during run-time every model respects Consumer Duty, SS1/24, the EU/UK AI acts and internal AI policies. CDOs must translate fine words into practical and enforceable guardrails before regulators, auditors or customers notice.
In a not-so-distant future geopolitical rifts could hypothetically throttle inter-continental data flows, raising latency, duplicating costs and stranding AI models trained on global datasets. CDOs must scenario-plan today: regional data zones, compliant encryption, local-first analytics and contractual exit clauses.
Imagine this:
Split into 3 teams – your scenarios are:
Each table’s nominated leader shares their team’s findings with the rest of the room
Data teams get excited by POCs, yet many fail to clear the final hurdles of risk and funding approval, as well as operational integration. As a result, time and money is wasted, innovative ideas get shelved, the board loses faith in data leadership and the data department once again gets branded as a high-expense cost centre. CDOs need to ensure they have clear strategies for POC selection and product implementation to create a better ROI for their business.
Chatbots trained on public large language models are prone to hallucinations, misquoting policies and leaking PII. RAG pairs a private LLM with permissioned document retrieval, serving proof auditors can trace. Leaders in the space are cutting call-centre average handling time while simultaneously slashing compliance redlines. It all sounds great – but CDOs and their teams need to establish robust architecture and governance, as well as evaluate the technology optimisation cost vs ROI.