A Seat, Not a Tab

Workiva released three AI agents last week that verify every number in a filing against its source, benchmark disclosure language against competitor SEC filings, and draft sustainability reports against ESRS and ISSB frameworks. These agents don’t sit in a chat window waiting to be asked. They run the close.

That distinction matters more than the technology. Most finance teams adopted AI the same way they adopted every other tool — bolted it onto the side and hoped someone would use it. Adoption hovered around 30 percent. Nobody could measure ROI because nobody changed the workflow.

Workiva changed the workflow. The tie-out agent checks every figure. The benchmarking agent runs comparisons as part of the process, not when someone remembers to ask. The sustainability agent drafts against two frameworks simultaneously with traceable citations. Each cycle feeds the next. The second quarter is better than the first.

This is the gap I keep watching widen. Companies deploying agents like tools — disconnected, siloed, invoked on demand — versus companies deploying agents like staff. Assigned work. Accountable output. A schedule. The average enterprise now runs 37 agents. Only a quarter have visibility into how those agents talk to each other.

The question was never whether AI belongs in the close. It’s whether your AI has a seat at the table or a tab in the browser. One compounds. The other collects dust.