Where do the other ten go? AI, reskilling and the revenue nobody sees

Where do the the other ten go? AI, reskilling and the revenue nobody sees

In June, two quite different organizations each released a distinct piece on AI. The first of the two, a report published by The Conference Board (TCB), tackled the subject of skilling the workforce for AI, while the second was Cognizant’s AI Forum, which included a presentation by WorkFabric AI’s CEO and co-founder Rohan Murty.

The TCB’s white paper entitled Skilling for AI: Critical Factors for Navigating AI Disruption is the result of interviews with 35 enterprise leaders and a survey of 1,266 workers globally. Like most surveys, it addresses several topics, but our piece centers on upskilling and reskilling.

Skills are at the heart of the AI debate. Specifically, what can employers and employees do to protect productivity and create new jobs where old ones disappear or need fewer people? A key point made by the TCB report is that AI skills training is often too basic; one where simple prompting techniques sit alongside AI literacy (e.g. What is an LLM?), and often completely omits training workers to use more advanced systems such as how to manage AI agents. In an era of tokeneconomics, the current approach does nothing to demonstrate how AI use can unlock real return on investment, or potentially turn areas previously considered cost centers into revenue generators in some capacity. More on this later.

The survey

The survey states that workers’ use of AI is outpacing formal training, with over half (55.1%) of employees using generative AI or AI agents daily or weekly. This contrasts with only 33.3% having used organization-provided training in the past six months (data below). Moreover, the results say, workers are not given enough time to learn AI skills, while employers argue that there simply isn’t enough time to train staff given existing workloads. What makes this challenge even more complicated is that with AI, training isn’t simply a course a member of staff has to pass; that is simply the basics and only part of a three-pronged approach that includes social and experiential learning recommended by the TCB.

This recommendation begs the question, if organizations are not even adequately tackling the easier task of upskilling, what happens when faced with the reskilling challenge in a fast-moving environment where there was only a 1.3-year difference between OpenAI releasing its first “reasoning model” and a leap forward in the quality of AI agents (Claude Code).

Figure 1

AI use is outpacing formal training

TotalEuropeUSAsia
Use generative AI daily or weekly49.4%48.1%47.6%56.6%
Used organization-provided AI training in last 6 months33.3%34.5%30.3%38.4%
Organization does not provide AI training28.3%20.6%33.9%30.5%
Q: How often do you intentionally use generative AI in your work?  |  Q: During the last 6 months, have you used any AI training resources your organization provides?
Note: Weighted results. Unweighted global bases: n=1,182 for generative AI usage; n=1,175 for organization-provided AI training.
Source: Skilling for AI: Critical Factors for Navigating AI Disruption, The Conference Board, June 2026

The reskilling challenge

Where training exists, it focuses on upskilling rather than reskilling for future needs. The report states that organizations will face ‘immense challenges when they need to reskill people as AI systems improve’. As an example, the report gives the view of a CHRO who said their organization is dealing with instances in which teams of 15 could be reduced to teams of 5. This statement is not an anomaly, and we increasingly see use cases where certain functions can be replaced with an individual managing a team of AI agents, should that individual have deep domain expertise. In the CHRO’s example, where do the other ten individuals go?

Part of that answer can be found in Cognizant’s recent AI Forum, in which Rohan Murty gave a demonstration of how AI is being used within the organization to unearth revenue opportunities that were previously unidentified. The use case uses signals from different teams (on-site, delivery, engagement) operating across multiple time zones (the US, the Philippines, India) to feed into the sales CRM to create new opportunities. By creating mission-driven twins, the company is able to generate net new revenue opportunities and, according to Cognizant’s CEO, has already identified USD 200mn in pipeline, which is expected to increase to USD 1bn by the year’s end. These account twins have also been used to prevent revenue churn, find more revenue likely to close in 120 days and even identify the right staff for certain programs based on who is already doing similar work.

Granted, the presentation focused on Cognizant’s ability to apply AI in order to drive additional revenue for clients, but the same approach could apply to much smaller entities, which have teams that equally operate in silos.

Our solution

As a workflow-automation consultancy, we support organizations by, identifying where valuable information is being lost between teams and systems, and building the workflows that capture it, structure it and route it to the relevant teams. We do this across multiple use cases and teams, though we often find the strongest demand from companies drowning in a lot of unstructured data that needs to be rationalized and turned into actionable data.

The value is not simply the time saved, though that is considerable. It is that the staff who once spent their hours monitoring and gathering are freed to move up the value chain into analysis and advice, which is where their domain expertise sits and where the return on investment is actually found.