Reskilling vs. Upskilling: It's Not a Course Catalog
A practical definition of reskilling in the AI era, and why most of what gets called reskilling isn't.
In September 2025, Accenture told investors it would cut more than 11,000 roles over three months as part of an $865 million restructuring, and its CEO put the new rules of the AI era about as plainly as anyone has. "We are exiting, on a compressed timeline, people where reskilling, based on our experience, is not a viable path for the skills we need," Julie Sweet said (CNBC). Reskilling is no longer a line in the benefits brochure. It is the test that decides who keeps a job.
It is worth being honest about why that sentence makes people flinch. In plenty of companies, "reskilling" is the polite word that arrives just before a layoff. But notice what Accenture actually did: it framed exits as the exception, not the plan. Sweet called upskilling its "primary strategy," and the company says it has trained more than 550,000 employees on the fundamentals of generative AI. Reskilling done badly is a euphemism. Done well, it is the actual alternative to cuts.
Reskilling vs. upskilling: what's the difference?
The two terms get used interchangeably, but they describe different jobs.
Upskilling means getting better at the role you already have. The analyst learns to drive a new AI tool. Same job, deeper.
Reskilling means moving someone into a meaningfully different role. The support lead becomes an implementation specialist. The operations analyst becomes the team's AI workflow lead. Different job, new core skills.
This distinction has real stakes, because most changing roles are a development problem rather than a turnover problem. The World Economic Forum estimates that of every 100 workers, 59 will need training by 2030. Of those, 29 can be upskilled in their current roles and 19 reskilled and redeployed elsewhere in the organization, leaving only about 11 likely to be left behind for lack of training (WEF Future of Jobs Report 2025). The implication is the one Sweet was gesturing at: develop people deliberately, and the large majority can be kept.
What real reskilling requires
Here is where the course-catalog approach falls down. Real reskilling has four features, and a generic library has none of them. To make them concrete, follow a single move through all four: turning a support lead into an implementation specialist.
Real reskilling is targeted. It aims at a specific gap between where the person is and a real role they could fill. For our support lead, that means naming exactly what an implementation specialist does that she does not do yet, rather than handing her a shelf of ten thousand courses to browse on her own time.
It is also built from your context. The most useful training teaches your tools, your customers and your processes, the way your company actually works. Our support lead needs to learn implementation on your product and your onboarding playbook, not a vendor's generic version of the topic. This matters more than ever for AI roles, because the useful knowledge often is not in any library yet. It lives in how your best people have figured out to use these tools on your actual work.
Crucially, the learning is active. People learn by doing. Reskilling our support lead means having her run a real implementation scenario with feedback, not autoplaying a video at double speed to move a progress bar.
Finally, it is verified. You should be able to tell whether the skill actually landed, not just whether the course was marked complete. Can she scope and run an implementation, scored against a clear rubric? Course completion and demonstrated capability are not the same thing.
The reason generic libraries fail is not that the content is bad. It is that browsing the world's knowledge will never be the same as building one specific capability, and a person staring at an endless catalog with no path will, reasonably, close the tab.
Where this gets practical
This is the gap Honen is built to close. Instead of pointing people at a library, it turns the knowledge you already have, your handbooks, SOPs, recorded calls and walkthroughs, into a structured course aimed at a specific role, in minutes. (Fast enough, in fact, that we have built several complete courses during a single 30-minute call.) Learners do not just read and watch. They work through hands-on scenarios and rubric-graded projects, with a live tutor that answers from your approved material rather than the open internet. That is the difference between handing someone a library card and handing them a path.
That is also what moves reskilling from a line in the annual report to something that actually happens, and it is what gives "we will reskill our people" the credibility to be more than a press release before a layoff.
The bottom line
Reskilling is a verb, not a subscription. It is the deliberate act of building a specific capability in a specific person, using what your company already knows, and confirming it worked. Everything else is theater with a completion rate attached.
In an era when reskilling is being named, out loud, as the thing that decides who stays, that distinction is not academic. If your reskilling strategy can be satisfied by buying a login, it is not a strategy. It is a library, and libraries do not change what your people can do.
Frequently asked questions
What is the difference between reskilling and upskilling?
Upskilling means getting better at the role you already have, the same job done deeper. Reskilling means moving into a meaningfully different role that requires new core skills, like a support lead becoming an implementation specialist.
Is reskilling the same as buying a course library?
No. A course library is a shelf of generic content. Real reskilling is targeted at a specific role, built from your own context, practiced actively with feedback, and verified by demonstrated skill rather than course completion.
Can most employees be reskilled, or is it cheaper to replace them?
The World Economic Forum estimates that of the workers who will need training by 2030, the large majority can be upskilled in place or reskilled and redeployed, and only about 11 in 100 are likely to be left behind. For most roles, development is the cheaper and more effective path.
How do you know if reskilling actually worked?
By measuring demonstrated mastery, not course completion. The person should be able to perform a real version of the new role's core task, scored against a clear rubric.
Want to see what real reskilling looks like, built from your own handbooks and recordings?
By StudyFetch Staff. Honen turns the materials your team already has into real courses they'll actually finish, built in minutes and measured by mastery.