AI Agents in Accounting: Practitioners Say 'None' Own a Workflow End-to-End — Here's What Actually Works

At a recent Earmark webinar on AI in accounting, practitioners were asked which accounting workflows AI agents fully own from start to finish without human review. The answer was equally direct: "None at all."

The July 8 session — "AI Agents for Accountants: What's Working Right Now" — brought together leaders from Earmark, Sweet, Karbon, and Gusto to assess the real state of agentic AI in finance. While interest is high, production deployments remain narrow and human oversight has stayed essential.

Key Takeaways

• Bain & Company research found only 7% of organizations have fully autonomous AI agents running in production today; Gartner has identified data context — not model capability — as the biggest barrier to reliable agent performance
• The most successful current AI deployments in accounting cover predictable, rules-based tasks: spreadsheet transformation, inbox management, dashboard generation, and recurring data formatting — not end-to-end workflow ownership
• Colleen Oates of Gusto captured the current reality: "75% to 80% of that workload is more efficient with AI" — but the remaining 20–25% still requires human judgment, especially in accounting close and client exception handling

Panelists consistently advised starting with automating repeatable, high-value workflows before expanding into more complex tasks. The session made clear that the gap between vendor hype and enterprise reality around agentic AI remains substantial — even among the practitioners actively building the technology.

Read the full article on CFO.com