More output isn’t better business: Workday's CFO Zane Rowe on how finance chiefs should really measure AI

Workday’s finance chief, Zane Rowe, is arguing that AI in the back office should be judged by what it changes, not by how many tasks it chews through. At the company, that thinking is already showing up in internal experiments, including a list of 10 AI agents the tax team is building to improve how employees work. Not all of them will survive. That’s kind of the point.

Rowe says Workday’s finance transformation team is working with research and development and business technology to test the company’s own products and poke at other tools too. The group also talks with customers to see how finance teams are actually using AI, which is often where the clean PowerPoint version of the story runs into the messy real world. Workday’s bet is that AI can do two things at once: sharpen business insight and redraw workflows from the ground up.

On the insight side, the company is expanding Decision Intelligence beyond Adaptive and into HR and finance. Rowe says the tool helps teams understand what’s driving performance, run scenarios, and estimate impact. That can mean modeling what happens if hiring changes in certain regions or business lines, then using those forecasts to guide planning. His point is blunt: this is not just forecasting with nicer branding.

Workday used its annual Rising event last week to announce new autopilot agents for routine accounting, payroll, and hiring work, plus a bigger Trusted Agent Platform for building and deploying custom agents. It also announced a new partnership with Anthropic to connect their platforms and agentic experiences. The timing matters. Workday is in what the company calls its “refounding” chapter, with co-founder Aneel Bhusri back as CEO since February.

Rowe’s bigger argument is that AI shouldn’t be scored like a factory meter. He says ROI still matters, but the real test is whether the work gets better: stronger judgment, better advice, more capacity, more value for customers. A team that can spit out twice as many reports is not automatically helping the business. The better question is whether it can find something useful inside all that output and help leaders act on it.