AI agents erase the paper trail, reshaping audit assurance

Enterprise audit teams are running into a strange problem: the work is still happening, but the trail it leaves behind is thinning out. Data remains in places like NetSuite, HR systems and data lakes. What’s disappearing are the little artifacts auditors used to rely on — email threads, Slack chats, handwritten notes and tick marks that showed how a decision got made.

That matters most in regulated businesses, where AI governance has moved far beyond a box-ticking exercise. At Workiva’s Amplify conference, that shift came through clearly. Agents can take over repetitive tasks and deliver real efficiency, but they also introduce a new kind of risk: the kind auditors can’t easily see.

Josh Robinson, chief audit executive at Vast Space LLC, put it bluntly. With people, the judgment was still visible in the record. With agents, that evidence can vanish. For him, the uncomfortable part is not that software is doing the work. It’s that the reasoning behind the work is becoming unknown.

The answer, Robinson argued, is not to invent some shiny new audit religion. It is to keep using the old discipline of completeness and accuracy, one step at a time, across the whole agent workflow. If those checks hold, auditors can still reach conclusions about the data or tests in front of them. But accountability does not move. Whoever signs the work owns it.

At Vast Space, that line carries extra weight because the company builds commercial space stations and faces astronaut safety risk as well as regulatory exposure. Robinson said that if something lands on his desk with his signature on it, it had better be correct. Efficiency matters, but so does control. Workiva’s pitch is that its AI agents and Knowledge intelligence layer can connect the scattered systems behind that work and help companies build agents across them.