AI Will Rewrite Every Workforce Plan · How to Join the 11% Who Already Factor It In

Only 11 percent of companies have built AI into their workforce and financial plans, a Conference Board survey released November 18 found. Here are three numbers a CEO should ask for, and four questions for the IT lead, before signing a 2027 headcount plan.
Three colleagues in business casual walk toward a brick office and plant building at golden hour, with a glowing blue overlay of linked people icons above the walkway.

Picture a $150 million manufacturer finalizing its 2027 plan. Department heads have asked for 24 new hires, which at an illustrative $95,000 each, fully loaded, comes to about $2.3 million a year. In the same building, people in customer service, finance and sales operations are already using AI tools, and nobody has measured what that changes. The CEO is about to sign a headcount number that says nothing about it. Whatever she decides, the board, the lender and the staff will read that number as her answer to what AI means for the company.

Budget season is usually about technology lines. Our piece on questions to ask before you approve next year’s IT budget covers those. This article is about the payroll lines, which are larger at most midsize companies and which almost nobody has tied to AI yet.

 

What Did 775 Leaders Say About Planning for AI?

On November 18, The Conference Board, a nonpartisan research group, released findings from a survey of 775 business leaders worldwide, backed by 35 interviews, on how companies plan their workforce as AI spreads through the business. Only 11 percent said they have built AI’s expected effects into both workforce and financial plans across the whole company. Another 35 percent have done it for some functions, roles, business units or regions. Thirty-four percent are discussing it with nothing in the plan yet, and 13 percent have not started to assess it. Add the last two and 47 percent have nothing in their plans.

The expectations behind those plans are mixed. Over the next two years, 39 percent expect a slight reduction in full-time employment, 8 percent expect a significant reduction, 32 percent expect little or no net effect, and 5 percent expect an increase. Seventeen percent say it is too early to estimate. On actions, about 70 percent expect redesigning jobs, tasks or workflows to be among their most impactful moves, and 68 percent name upskilling. Thirty-seven percent plan to invest in AI capability instead of adding headcount, and 17 percent name cutting full-time headcount in AI-affected areas as a top action.

Read the limits before you quote it. The public release does not say when the survey was fielded, and it does not break respondents out by industry or company size, so these leaders are not necessarily running midsize companies. They are describing intentions, not results. The full report is open only to members, and we worked from the public release. What the numbers support is narrower and still useful: nearly half of the leaders surveyed have not yet written down what AI changes in their own plans, so a CEO who does it this winter is ahead of that half, not behind the field.

What the numbers support is narrower and still useful: nearly half of the leaders surveyed have not yet written down what AI changes in their own plans, so a CEO who does it this winter is ahead of that half, not behind the field.

 

Why Can’t One Headcount Number Answer This?

Because the effect of AI rarely arrives as a clean cut. Matt Rosenbaum, a principal researcher at The Conference Board, said in the release that “Eliminating one task can create new responsibilities elsewhere.” A support team that no longer writes first replies by hand may now spend that time reviewing drafts, handling the harder cases that used to be buried in the queue, and correcting errors. Time saved does not turn into fewer people unless someone decides it will and redesigns the work around that decision.

Time saved does not turn into fewer people unless someone decides it will and redesigns the work around that decision.

That is why the survey’s most common planned action is redesigning jobs and workflows, not cutting. It is also why a flat growth assumption and a flat freeze can both be wrong. Freeze every requisition and you may starve a team whose workload is growing faster than any tool can absorb. Approve every requisition and you may fund 2026’s way of working for another year, then discover in the third quarter that the work changed and the payroll did not. In our picture, the CEO has a number she cannot defend against either question.

 

What Three Numbers Should You Ask For Before You Sign?

First, the hours. Ask each department head how many hours a week of recurring work has actually changed because of an AI tool, measured over the last 60 days, and in which roles. Not what a vendor promised. A rough count from the team lead, with the method written down, is enough to start. If the answer is that nobody knows, that is still an answer, and it belongs in the plan as a gap with an owner and a date.

If the answer is that nobody knows, that is still an answer, and it belongs in the plan as a gap with an owner and a date.

Second, the hires you will not make. For each of the 24 requested positions in our picture, ask whether the request exists because the work is growing or because the work is the same and the team is stretched. A request that exists because the team is stretched is the one an AI tool might cover. Pair each such request with a trial: fund the tool, hold the hire for one quarter, and decide on evidence. That is the 37 percent approach in the survey, buying AI capability instead of adding headcount, and it works only if the saved hire is written into the plan next to the tool’s full cost, including the usage charges we described in our piece on AI tools that bill by the click.

Third, where the money goes. If an AI tool does give back time or cost, someone must decide where it lands: margin, growth hiring somewhere else, training, or a reduction. The Conference Board’s September report on agentic AI and work redesign ends its five-stage framework on the same question, which is where the gains go. Left undecided, the gain is quietly absorbed as slack. We argued in our piece on a role-model layoff that how a company answers this question becomes part of its reputation with its own people.

 

What Will Your Board, Lender and Staff Ask?

In our view the board’s question will be short: what in the 2027 plan changes because of AI, and how will we know? Our earlier piece on what to tell a board that asks if you are behind gives the opening answer. A lender reading a budget with growing payroll and a separate line for AI spending will ask the same thing from the other side: is the company paying twice for the same work? A buyer or investor doing diligence will ask which savings are already in the run rate and which are hopes.

Staff ask a different question, which is whether the plan means their job. The survey shows why silence is costly: 47 percent of leaders, by our sum of the first two figures, expect fewer full-time employees within two years, and your employees read the same headlines. A CEO who says what the plan is, including what has not been decided about roles, earns more trust than one who lets a rumor set the number. If you want a board-level view of who should carry this question, see why your board needs an AI strategist.

 

What Should You Ask Your IT Lead?

You do not need the technology explained. You need four answers on one page, in dollars and hours, from someone who will be asked to repeat them to the board. Which AI tools are in daily use in the company, including ones staff signed up for on their own? For each, which team uses it and for what work? What does it cost per month today, and what happens to the bill if usage doubles? And what can we measure now, such as handling time or turnaround, that we could not measure a year ago?

 

What This Will Not Do

A workforce plan will not tell you the right headcount. The survey measures what leaders intend, not what worked, and its sample is not built around midsize companies. Our judgment in this piece, that a flat number can mislead in both directions and that a one-quarter trial is a safe way to learn, is ours and not the survey’s. Nothing here is a reason to cut people or to avoid cutting them. Reductions, changes in roles and notice rules depend on your state, your contracts and your people, so take them to employment counsel. This is not legal advice.

 

Where to Start

Start with one page before the budget is signed: for each open requisition, the work it exists to do, and whether an AI tool already touches that work. Fifteen minutes with each department head is enough. If the page shows you cannot say, that is the finding, and it is the cheapest one you will get. If you want a second set of eyes on what your teams are using and what it costs, our assessments start from exactly that inventory, and CIO IQ gives you an experienced CIO’s advice at $1k per week when you want the plan pressure-tested before it goes to the board. You can reach us through the contact form on our about page.

More from our blog

AI Will Rewrite Every Workforce Plan · How to Join the 11% Who Already Factor It In

AI Will Rewrite Every Workforce Plan · How to Join the 11% Who Already Factor It In

Only 11 percent of companies have built AI into their workforce and financial plans, a Conference Board survey released November…
Governance Starts with Visibility · Corporate AI Monitoring in 2027

Governance Starts with Visibility · Corporate AI Monitoring in 2027

Whether you can show what your people did with AI depends on the plan your company bought. We compare what…
Who Vouches for Your AI? · Governance at the Leading Edge

Who Vouches for Your AI? · Governance at the Leading Edge

Boards, examiners and buyers do not trust software that checks itself. They trust accountable people who test, sign and keep…