AI agents complicate audit trails as software outpaces human oversight

Auditors across enterprises are confronting a troubling reality: when AI agents take over operational decisions, the documentation that has long anchored financial oversight vanishes. While raw data continues to flow through NetSuite, HR systems and data warehouses, the reasoning behind critical choices—once captured in email chains, Slack conversations and handwritten notes—now occurs within software that moves at speeds no human can track.

The problem cuts deepest in regulated sectors, where AI governance has climbed from a compliance formality to a board-level priority. This tension defined the conversation at Workiva Inc.'s Amplify conference, where speakers grappled with how to extract value from AI agents while managing the risks they introduce. Josh Robinson, chief audit executive at Vast Space LLC, frames the core challenge starkly: auditors have lost visibility into the very reasoning they need to verify.

With humans, we could understand they were using their judgment, they were exercising probability, but we could see it. There was either handwritten notes or there were tick marks or there was email chain or Slack messages. That has all been removed from the equation. So, for me, the risk is in the unknown.

Josh Robinson, chief audit executive at Vast Space LLC

Applying traditional audit discipline to algorithmic workflows

The solution does not require inventing new processes; instead, it demands applying time-tested audit principles to AI-driven operations. Robinson explains that internal audit teams have long relied on two foundational tests: completeness and accuracy. These same standards can be applied methodically across each stage of an agent's workflow.

If you can demonstrate completeness and accuracy as an auditor, generally that allows you to reach conclusions about the assurance of datasets or audit tests you're doing.

Josh Robinson

Accountability remains the harder problem to solve. At Vast Space, which manufactures commercial space stations and manages astronaut safety alongside regulatory obligations, control failures carry consequences far beyond financial restatements. The person whose signature appears on the work retains ultimate responsibility, regardless of whether a human or an algorithm produced it.

As an audit leader, if it comes to my desk and I'm going to put my signature on it, it pretty darn well better be the right answer.

Josh Robinson

Breaking down data silos to enable AI coordination

Extracting maximum value from AI agents first requires consolidating fragmented data sources. Workiva's approach combines specialized AI agents with a Knowledge intelligence layer designed to bridge the gap between disconnected systems. Robinson, who has been a Workiva customer since 2017, sees the appeal in a single integration layer that can connect NetSuite, HR platforms, data lakes and whichever AI model an organization has deployed.

https://www.youtube.com/embed/9PxqviEEBeM?feature=oembed

I can go to my CFO and be like, OK, we use the likes of a NetSuite and an HRIS and this and this, and this is our data lake. And we're using Claude, but now look at Workiva who can connect to those 5 different things, bring it in and create agents simply that is going to solve X, Y, Z problem for you.

Josh Robinson