At Risk Oversight, much of our work is advising CFOs and Controllers on governance, risk, and compliance: internal controls, process design, and finance transformation. They’re some of the hardest-working people I know, and the value sitting inside their work is often far greater than their organizations realize. So in this issue of Leverage Your Knowledge, I’m focusing on one under-used source of that leverage, one that never shows up on a controls matrix: imagination, and the willingness to pick up the tools now within reach.
To my CFO, Controller, and accountant friends, humor me for a minute.
The finance leaders who thrive in this era won’t be the ones with the biggest tech budgets. They’ll be the ones with the most imagination, and the willingness to change something once they’ve imagined it.
That second part matters. Imagination without follow-through is just a pleasant afternoon. But you can’t follow through on something you’ve never let yourself picture. So let’s start there.
The real story is the price
What keeps catching my attention isn’t what AI tools can do. It’s how cheap they’ve become. This isn’t an Oracle, SAP, or Workday implementation, with a steering committee, a six-figure statement of work, and a go-live date that slips twice. It’s automated journal entries running on Claude or ChatGPT for less than a steakhouse dinner.
Remember when OCR (optical character recognition, mostly reading data off invoices) was the hot new thing in accounts payable? It cost a fortune and took a year to roll out. Today it’s a $30 subscription with almost no setup. The technology improved, but the real shift is that the price fell from a capital project to a line item you’d lose in your coffee budget.
My New Year’s Eve with a robot bookkeeper
About a year ago, I started using Claude for my own bookkeeping. It’s my least favorite task on earth, which is exactly why I save it for New Year’s Eve, like the party animal I am. I expected the usual grind. Instead, it categorized, reconciled, and flagged the odd items faster than I could pour a glass of bubbly. When I tell finance professionals this, I usually get a nervous laugh, then: “Wait, it can do that?”
Yes. And it’s about to do a lot more.
The next leap is agentic AI, where the tool doesn’t just suggest the work, it does it. An agent books the entry, runs the reconciliation, drafts the variance commentary, and hands you something to review rather than build. You move from doing the work to checking it, which is closer to what we already say a controller’s job is.
“But is it safe?”
Before the CFOs reach for the security flag: both ChatGPT and Claude hold SOC 2 Type II and ISO 27001 certifications, the same standards behind your accounting, tax, and audit software, with independent testing every year. (You can see them for yourself on the Anthropic Trust Center and the OpenAI Trust Portal.)
It’s still a fair question. I’ve read the rogue OpenClaw stories too, including the one where Meta’s own AI alignment director told her agent to “confirm before acting” and watched it start mass-deleting her inbox while she typed “STOP” into her phone. But the pattern is almost always the same: a half-baked consumer tool handed the keys with no guardrails, not the enterprise software your team would actually run. The fear is usually overblown, and too much of it will quietly put you behind.
Wall Street has placed its bet
In May, Anthropic launched a suite of AI agents built for financial work: pitchbooks, audited statements, credit memos, compliance checks. CEO Dario Amodei shared the stage with JPMorgan’s Jamie Dimon to announce it, and major players like Goldman Sachs and AIG are already on board. The tools plug into Excel, PowerPoint, and Outlook and pull live data from sources like Moody’s and Dun & Bradstreet.
Here’s the part I’d circle. The day before that launch, Anthropic announced a $1.5 billion joint venture with Blackstone, Goldman Sachs, and others, aimed at the mid-market and embedding these tools directly in the operations of mid-sized companies. This isn’t a big-bank toy. It’s coming for the finance function of the company down the street, and the timeline is shorter than most of us are planning for.
Yes, it’s coming for the Big Four too
That same venture has been described as a direct shot at the consulting industry: a turnkey alternative to the Big Four at a fraction of the cost. The pyramid model that built these firms is under pressure, and the firms know it. Just a couple of weeks later, KPMG announced a global alliance to put Claude in front of all 276,000 of its people across 138 countries, and to become Anthropic’s preferred partner for deploying AI in private-equity portfolio companies. If you can’t beat them, join them.
Imagination needs fuel
Here’s the catch. Press releases and thought pieces (this one included) won’t get you there. Real learning comes from getting your hands dirty: opening the tools and trying them on your own ledger and your own mess. But you can’t experiment toward something you can’t picture. Your imagination needs raw material first: examples, demos, something concrete to react to. Before you can play, you have to watch.
Where? YouTube, of all places. It’s not just cat videos and whatever your teenager has on loop; it’s become one of the best windows into what’s possible in our profession. (Yes, I have a channel). The subscriber count is what auditors would politely call “immaterial.”)
There isn’t a great channel for CFOs yet, which may be an opening for one of us. The one I keep returning to is Jason on Firms by Jason Staats, CPA. He demos the unglamorous work that eats your team’s week: journal entries, tax workflows, bookkeeping, financial analysis. His videos are well produced and funny, but he’s speaking to small-firm owners, and his examples are sized accordingly. Watch a few anyway. You’re not looking for a blueprint to drop into your enterprise. You’re watching the mechanics up close and letting your imagination scale them up. Expanding your imagination usually starts with expanding who you follow, beyond people in your exact seat.
“Isn’t that my team’s job?”
A CFO friend and client asked me that recently. Fair question. But you’re the one signing off on headcount, consultants, process design, and training budgets. If you don’t know what’s possible, you’ll keep buying yesterday’s solution at tomorrow’s price, and approving the workaround instead of asking whether the work needs doing at all. The risk isn’t that AI replaces finance leaders. It’s that finance leaders who can’t picture it keep authorizing work that no longer needs to exist.
To be clear, I’m not suggesting you rip out SAP and run your close and financial reporting on ChatGPT. I’m not suggesting we discard the structure, judgment, and controls the profession has built since Luca Pacioli codified double-entry bookkeeping in 1494. And I’m certainly not suggesting you replace your people with bots.
I’m just curious what happens when finance leaders let their imaginations off the leash for one afternoon. What if you asked:
- Where is there still manual entry no one has questioned in five years?
- Which clunky process does everyone quietly tolerate?
- Where do you need another set of eyes but can’t justify the headcount?
- Are your controls designed for 2026, or stuck in 2015?
- Are you feeding your imagination with demos and examples that are bold but realistic?
- And what practical first step would move one of these ideas forward?
The CFOs and Controllers who win this era will be the ones who can look past the rough edges of a YouTuber’s demo, the confines of the big-firm model, and the comfort of “the way we’ve always done it,” and ask the real question: what could we do with tools this cheap, this capable, and this trusted?
Imagination might just be the modern CFO’s power skill. Worth a try, anyway.