Workiva Amplify: traceable AI agents needed for finance

The discussion at Workiva's Amplify event, held during an exclusive broadcast on theCUBE, SiliconANGLE Media's livestreaming studio, framed AI governance as a workflow problem rather than a purely technical one. Workiva operates in reporting, audit and compliance workflows where errors carry serious consequences, which is why the conversation focused on control and accountability rather than model performance. Case, speaking with host Alison Kosik, tied the challenge to AI's reach: because AI can process information faster and across more workflows, it also magnifies weaknesses that predate generative AI.

Those weaknesses — fragmented data stores, inconsistent definitions and inconsistent ownership — are, in Case's account, not new problems created by enterprise AI adoption. What has changed is urgency: agents can turn flawed information into decisions at greater speed and scale. That puts the emphasis on controls built around risk, with monitoring and exception handling becoming more important as use cases expand.

A central open question is where the line between assistance and autonomy falls. Case suggested those boundaries are still being defined and will evolve, particularly as business use cases evolve, which leaves companies weighing efficiency against oversight. How strictly firms require human approval for agent actions — and whether they can reconstruct those actions after the fact — is likely to shape how quickly AI agents are trusted with consequential work.