From accidental fire alarms to the rise of "Agentic AI," the 2nd Annual AI for CFOs summit moved the needle from theoretical hype to hard-nosed pragmatic deployment. Here is what the modern finance leader needs to know about the next 12 months.
This isn’t just another tech conference; it’s a reality check. The 2nd Annual AI for CFOs summit, hosted by The Economist, kicked off with a literal fire alarm. It was an accidental but perfect metaphor for the current state of finance: there’s a lot of noise, a sense of urgency, and everyone is trying to figure out if they should head for the exits or lean into the heat.
Last year was about the “wow” factor of Generative AI. This year, the theme was “Risk, Reward, and Reinvention,” and the conversation shifted from philosophical “what ifs” to the gritty, pragmatic “how-tos” of deployment.
For the CFOs in the room, representing heavyweights like Radisson Hotel Group, Bayer, and Danmarks Nationalbank, the consensus was clear: the honeymoon phase with AI is over. We are moving out of isolation and into enterprise-wide implementation.
Konstantin Dzhengozov, CFO of Payhawk, dropped a sobering stat: while a survey of 1,500 finance professionals showed high enthusiasm, fewer than 50% have actually automated a single process with AI. The “noise” is real, and the challenge for the modern CFO is cutting through it to find actual economic value.
The breakout star of the day wasn’t a person, but a concept: Agentic AI. As Rajesh Jindal, CFO EMEA at Amazon Web Services (AWS), explained, we are moving beyond traditional “if/then” automation.
Traditional AI flags an unusual expense; Agentic AI investigates the variance, pulls data from inconsistent sources like newspaper articles or HR systems, synthesizes an analysis, and then critically schedules the follow-up emails to the relevant teams. It’s not just a tool; it’s an autonomous worker.
One of the most relatable sessions tackled the “CFO’s Verdict” on value. The struggle isn’t just in the tech; it’s in the Excel lines.
Jeremiah Crider of Carwow pointed out a fascinating tension in tracking ROI. When AI identifies a revenue leakage or a cost saving, the “human” regions often try to negotiate a win-win. “I realized that leakage even without the tool,” they might say. The lesson? CFOs must be ruthlessly objective about KPIs and track usage as much as revenue. If a tool is sitting on the shelf, it’s a cost, not an asset.
In a refreshing moment of candor, the speakers didn’t just share success stories; they talked about the projects that hit a wall. Stephanie Rinkel from Bayer emphasized that organizations must facilitate “mistakes as a learning process”.
The common thread in failures? Trying to automate “chaos”. If the underlying data “plumbing” is messy, AI won’t fix it, it will just make the mess faster. The advice from the stage: focus on the “Sweet Spots”, high visibility, high business value areas like advanced forecasting and reporting and leave the “Quick Wins” to the junior teams to experiment with.
Perhaps the most important takeaway for every US and UK CFO in the room was the “Human in the Loop” principle. Graeme Fleming (Workiva) offered a perfect analogy: treat AI like a high-potential junior analyst.
You wouldn’t let a junior analyst publish a board report or make a strategic pivot unsupervised; you review their work, challenge the “black box,” and add the business context that the machine lacks. As Riccardo Calliano (GSK) noted, AI optimizes a single function, but humans make choices weighing values, long-term objectives, and organizational impact.
The fire alarm at the start was wrong, the building wasn’t burning down, it was just waking up. AI isn’t here to take our jobs; it’s here to make us “architects of systems” rather than processors of data.
The roadmap for the next 12 months?
As the summit concluded, the message was clear: AI won’t replace managers, but managers who use AI will replace those who don’t. See you all at the next close.