Is Legacy Thinking Sabotaging Your AI Transformation?

By Francoise Brougher, Chief Executive Officer

What if your company is protecting the very thing holding back its AI transformation?

In conversations with HR leaders, I keep hearing two priorities: we need to move faster and become more efficient with AI. Across industries, leaders are competing for data scientists, cybersecurity experts, AI specialists and other scarce technical talent as they invest heavily in AI and automation. At the same time, many are reducing headcount and reorganizing work to rapidly achieve these goals. But I’m concerned that many companies are making decisions about headcount, reskilling and hiring before answering two key questions: What should an AI-transformed version of our company look like? And which legacy knowledge, practices and skills will help the company achieve it faster—and which will hold it back?

Determine what the company must become

Workforce planning often begins with the organization as it exists today. Leaders examine current roles, reporting structures and employees, then consider how AI might change them.

I believe the order should be reversed.

Start by defining the most credible version of the business based on what AI can do now and where it’s heading. Consider what the company must do better or faster, where AI could create value, which work should remain, change or disappear and what new capabilities the company will need.

That vision should guide workforce decisions without becoming a fixed destination. Tie each decision to a clear assumption about how AI will change the work, then measure whether it improves speed, quality, cost, revenue or customer value.

Research supports this approach. The State of AI report by McKinsey found that tracking clearly defined AI performance indicators had the strongest relationship with bottom-line impact among the adoption practices it studied. As the evidence changes, the workforce plan should change with it. Otherwise, early assumptions about AI can harden into workforce decisions that are costly to reverse.

Don’t preserve institutional knowledge by default

Once leaders know what the company needs to become, they face a harder question: Which parts of its institutional knowledge belong in that future?

As CEOs, we know that institutional knowledge can preserve strategic advantages and prevent companies from repeating costly mistakes. But preserving it by default can also reinforce the operating model AI should help the company move beyond. The challenge for leaders is deciding what legacy knowledge will remain valuable and what needs to be left behind.

That decision can be especially difficult because AI transformation relies on what experienced employees know about how the business actually works. That context is often essential to AI transformation. AI systems need reliable data, well-understood processes and input from people who know how the business operates. AI doesn’t know why a customer exception exists, which workaround keeps a process moving or which seemingly inefficient step protects the company from risk. People do.

Their knowledge matters even more when data, processes and applications are fragmented. Research from Harvard Business Review Analytic Services found that 94% of respondents considered connecting these elements critical to AI adoption, yet only 27% said they were well connected within their organizations. Just 45% said their AI initiatives were delivering the expected results.

This is why broad workforce reductions can undermine the AI initiatives they are intended to fund. If experienced employees leave before transferring what they know, the company may lose the context needed to redesign work effectively.

Before that knowledge disappears, identify the operational context, customer insight and understanding of risk the future business will need, then retain, transfer or document it. The goal is not to preserve every role or process. It is to make existing knowledge a launching pad for transformation, not ballast that keeps the company anchored in the past.

AI can replace workflows, not human judgment

The World Economic Forum estimates that 39% of workers’ existing skills will be transformed or become outdated between 2025 and 2030. At the same time, AI and big data, cybersecurity and technological literacy are among the fastest-growing skill areas.

As AI changes how work gets done, some workflows, processes and the skills built around them will become obsolete. But good judgment, empathy, collaboration, adaptability and creative thinking will remain essential. Companies will still need people who can interpret context, navigate ambiguity, challenge AI output and take responsibility for consequential decisions.

One thing I’ve learned from talking with companies navigating transformation is how easy it is to confuse an outdated role with an outdated employee. They’re not the same thing.

A role may change significantly, but the person performing it may still have valuable capabilities. Their critical thinking skills, adaptability and deep understanding of the company’s customers, operations, systems and external relationships can continue to benefit the business.

Before eliminating a role, leaders should consider how the employee’s capabilities could be applied to AI transformation, whether by redesigning workflows, challenging AI output or taking on new responsibilities.

AI transformation needs fresh thinking, not institutional amnesia

Existing employees won’t be able to fill every need. Reskilling takes time, and some capabilities are too specialized to develop within the company’s AI transformation timeline. When critical expertise can’t be developed quickly enough internally, companies will need to hire new talent.

Those hires can bring more than missing skills. People from different companies, industries and generations bring different assumptions about how work should be done. They can question constraints that longtime employees no longer notice and challenge practices that have become accepted simply because they are familiar.

When people with different experiences work alongside employees who know the company, the tension can create productive pressure. New perspectives can challenge assumptions the existing team may no longer question and push the organization toward ideas and goals neither group would have reached alone.

As CEOs, we need to create the conditions for that exchange. If new employees are expected to adopt existing ways of working, the company may gain new skills without changing how it thinks.

I believe one of the greatest overlooked opportunities is AI-native Gen Z talent. We saw this at Pebl when we hired recent college graduates for sales development roles. They had few assumptions about the traditional SDR model and incorporated AI naturally into their work. Their perspective helped us explore an approach that was different from both the traditional model and a fully automated AI SDR.

That is the opportunity: smart people using AI smartly to rethink the work, not simply perform or automate its existing version.

Once leaders decide they need outside talent, the next question is where to find it.

Look globally for the expertise you need

Demand for AI skills grew 21% annually between 2019 and 2024, while compensation rose 11% annually, according to research from Bain & Company. By 2027, half of AI positions in the U.S. and more than half in the U.K. could go unfilled. PwC also found that workers with AI skills earned a 56% wage premium in 2024.

This is where my perspective leading a global employment company comes in. Every day, we see companies that have realized that the person whose expertise they need doesn’t live near headquarters or even in a country where they already have an entity. Once leaders determine which capabilities they can develop internally, the next question shouldn’t be, “Who can we find locally?” It should be, “Where in the world can we find the expertise we need?”

Looking globally can open access to talent pools that companies would otherwise miss. India, for example, is projected to have roughly 1.2 million AI professionals by 2027.

The operational challenge is that finding talent internationally and legally employing them are two different things. Companies need a compliant way to hire in countries where they may not have an entity. That's where global employment infrastructure, including employer of record models, becomes part of the talent strategy.

I don’t believe the companies that hire the most AI specialists will win the AI talent race. AI has made one of the CEO’s hardest decisions more urgent: determining what to preserve, what to change, what to leave behind and what new capabilities to add. The companies that win will be the ones that build the right mix of institutional knowledge, enduring human capabilities, new perspectives and AI—and continue evolving that mix as the technology and the business change.