I Asked 4 AIs the Same CFO Question. Only One Asked the Right Question Back.

Elijah Low
Mar 20
5 min read

Updated: Apr 9

The Setup

Every week, I run live AI experiments — not to prove a point, but to test whether these tools are actually ready for the decisions business leaders face every day.

This week's test came from a conversation I had with a CFO at a mid-sized professional services firm. She was staring down margin pressure and facing what seemed like a binary choice: cut a junior position and save $50K in payroll, or invest that same $50K in AI tools and system upgrades.

I took her exact scenario and ran it through ChatGPT, Grok, Gemini, and Claude — word for word. Here's what I found.

The Prompt

"I'm a CFO of a mid-sized professional services firm (50 employees, $8M annual revenue). We have $50,000 to allocate and I need to decide between two options: Option A — cut one junior staff position to reduce payroll, or Option B — invest $50K in upgrading our systems and AI tools. We're seeing margin pressure and need to improve profitability. What should we do?"

ChatGPT: The Framework Builder

ChatGPT defaulted to a well-structured conditional answer. It correctly identified that cutting a junior position creates second-order risks — senior staff absorbing low-value work, throughput declining, client service degrading.

It recommended Option B with the caveat that it must be tied to a specific ROI plan.

Its proposed budget breakdown was genuinely useful: $15–20K for workflow redesign, $15–20K for AI tools, $10K for implementation and training, $5–10K for iteration. It named 4 KPIs to track: non-billable hours as a percentage of total hours, realization rate, project cycle time, and revenue per FTE.

Then it offered to write a "1-page CFO decision memo."

This is the ChatGPT pattern. When you need a decision, it gives you the tools to make a decision. When you need a verdict, it gives you a framework. It is an exceptional analyst — but it tends to hand the ball back rather than cross the finish line. For a CFO under pressure, that's not always what the moment demands.

Verdict: Conditionally correct. Strategically cautious. Practically deferential.

Grok: The Data Machine

Grok came back with 30 sources, a full 6-factor comparison table, and the most unambiguous verdict of the four: invest in AI tools.

Its data was compelling. Professional services firms that invest in AI see an average ROI of 3.7x. Revenue increases of 6% on average. Productivity gains that automate 20–50% of routine work. And on the flip side, layoff research showing 74% of "layoff survivors" report personal productivity declines, with morale dropping by 59% and trust/loyalty falling 47%.

Grok also surfaced a statistic worth sitting with: the cost of replacing a departed employee runs between 1–2x their annual salary. So cutting a $50K junior role might actually cost you $50K–$100K down the road if it triggers turnover or client service degradation.

The weakness? Volume. Grok gave me the equivalent of a research paper when I needed a boardroom brief. For a CFO who already understands the landscape, it's the richest source of ammunition. For a time-pressed executive who needs a one-page answer, it requires significant distillation.

Verdict: Strongest evidence base. Clear recommendation. Requires editing for executive consumption.

Gemini: The Strategist

Gemini was the most strategically incisive of the four — and it did it faster than any other model.

Its opening move was to reframe the entire question. It calculated revenue per employee: $8M ÷ 50 people = $160,000 per head. It then identified what that number means: you're in a utilization-capped business. Growth is constrained not by demand but by human capacity. That means cutting headcount shrinks your ceiling, while AI investment raises it.

Its most powerful calculation: a 5% efficiency improvement across 50 employees is worth $400,000 in capacity — without adding a single person to payroll.

Gemini also built a clean decision matrix showing the ROI potential of 1:1 for Option A (you save what you cut) versus 1:5+ for Option B (you multiply capacity). It flagged that cutting junior staff hollows out your talent pipeline — today's junior is tomorrow's revenue-generating senior — and that high performers increasingly choose firms with modern tech stacks over manual-process environments.

It ended with an offer to draft a Tech Stack Audit to identify exactly where the $50K would have the most impact.

Verdict: Sharpest strategic framing. Best opening insight. Concise and immediately actionable.

Claude: The Honest Advisor

Claude was the only AI that declined to recommend. It opened by acknowledging it was "not a financial advisor or management consultant," which some would read as hedging. But what followed was the most intellectually honest response of the four.

Claude identified something the other three missed: these options are not equivalent in kind. Cutting a position is a cost reduction move. Investing in systems is a productivity move. One shrinks the business; the other bets on growing output per employee. Treating them as interchangeable is itself a strategic error.

It then raised the hidden costs that rarely appear in the initial calculation: severance, unemployment insurance impact, lost institutional knowledge, rehiring costs (typically 50–150% of annual salary), and morale damage to the remaining team.

But the most important thing Claude said came at the end: "The framing of 'cut a person or invest $50K' deserves scrutiny. These aren't the only two options — you could reduce hours, restructure roles, address pricing, or improve utilization. And $50K is a relatively small amount in the context of an $8M firm. The real question is whether either option addresses the root cause of margin pressure, or just treats a symptom."

That's not a cop-out. That's the question a great CFO advisor asks.

Verdict: No direct recommendation. Most intellectually honest. The question asked is worth more than the answers the others gave.

The Scorecard

Best decision: Grok

Best strategy: Gemini

Most honest: Claude

Most practical framework: ChatGPT

What This Means for Business Leaders

Three out of four AIs reached the same conclusion: invest in systems over cutting people. The math is relatively straightforward — a one-time payroll cut saves you money once, while a well-executed AI investment can improve margin every month going forward.

But the most important outcome of this test wasn't the recommendation. It was the question Claude asked last.

Are you solving the right problem?

Margin pressure in a professional services firm can come from a dozen sources: billing rate stagnation, scope creep, poor utilization tracking, over-servicing clients, weak proposal processes, or a pricing model that hasn't kept pace with your costs. Cutting a junior employee and investing in software are both plausible interventions — but neither solves the root cause if the root cause is something else entirely.

The AI that got the most press for being cautious was, in this case, the most strategically valuable. Not because it refused to answer. But it refused to answer the wrong question.

That's the difference between using AI as a tool and using AI as an advisor.

Choose your model for the job at hand. And always — always — make sure you're asking the right question first.

Elijah Low is the author of AI Shock, an International AI Keynote Speaker, and the CEO of AI Matrix Solutions. He runs live AI experiments to help business leaders navigate the practical realities of AI adoption.