Risk and Cost Governance for AI Agents in Regulated Institutions - with Shahir Daya of Zafin
The AI in Business PodcastShow Notes
As AI agents move from pilots into live financial workflows, most institutions still can't answer a basic question: who authorized this action, and what evidence backs it up?
In this episode, Shahir Daya, Chief Product & Technology Officer at Zafin, examines why uniform governance policies fail at scale and what a tiered, control-tower approach to agent oversight looks like in practice. The conversation covers execution-layer governance, cost visibility across models, and the shift required to move agentic work from promising pilot to defensible production.
This episode is sponsored by Zafin.
Learn the exact strategies we use to help leading AI brands and startups connect with their ideal enterprise AI buyers: visit emerj.com/AD1
More from The AI in Business Podcast
Service organizations hold most of their critical expertise in experienced technicians rather than manuals, and that knowledge is leaving as the workforce turns over. In this episode, Thomas Cottereau, CEO and Founder at SightCall, explores why the real challenge is capturing expertise rather than producing more documentation, and how AI can structure what experts already know into trusted knowledge. He outlines how to capture knowledge inside everyday service work, keep human experts reviewing what AI produces, and treat that knowledge as a continuous loop that feeds both technicians and existing AI systems. This episode is sponsored by SightCall To learn how leading AI brands like NVIDIA and Google Cloud work with Emerj to reach Fortune 500 AI buyers - download our media kit at: emerj.com/AD1
Enterprise investment in AI is rising faster than the ability to prove what it returns, leaving technology and vendor spend largely unexamined until leadership asks whether it is working. In this episode, Kellen Winters, Director of Customer Success and Business Operations for AI at Novartis, examines why demonstrating AI value starts with building a measurement discipline before any tools are chosen. The conversation covers anchoring AI priorities in the processes where work stalls or drains resources, narrowing a long list of use cases to the few that can be measured, tying pilots to existing KPIs within 90 days, and using a structured intake process to track value from the first idea onward.
A job title no longer tells you what a technical role actually requires, now that AI has changed how the work gets done. In this episode, Carrol Chang, CEO of Andela, unpacks how AI is reshaping technical roles, and why hiring and assessment need to shift from job titles to skills.
Shifting regulations and siloed risk and compliance functions are forcing enterprise leaders to rethink audit-driven compliance. In this episode, Craig Unger, CEO and Founder at Hyperproof, explains why compliance should be treated as a continuous state, with every control tied back to the organization's overall risk profile. He also covers standardizing a single set of controls across frameworks and regions, expressing risk in financial terms for boards, and using a five-stage maturity model to plan next steps.
Financial institutions are moving fast on generative AI without a clear answer for who owns, and who is liable for, the content it produces and consumes. In this episode, Stefan Zima, Head of Group Regulatory Compliance and AI Compliance at Raiffeisen Bank International AG, examines how banks can build real accountability and visibility into their AI content practices before regulators or auditors force the issue. The conversation covers cross-border data governance, third-party AI risk, and where research reports, market commentary, and training materials create the most exposure. Learn how leading AI brands like NVIDIA and Google Cloud work with Emerj to reach Fortune 500 AI buyers - download our media kit at: emerj.com/AD1