AI taking over work does not only happen internally. A large part of the shift comes through the suppliers you already work with: the CRM package that suddenly offers a module that writes reports itself, the accounting software that processes invoices without human review, the customer service tool that handles tickets without an employee looking at them. The assumption underlying this is usually simple: "this work is done by us, or one of our people, or a party we hire." As soon as a supplier builds that step into the product itself, that assumption is no longer true without a decision having preceded it.
The signal is not found in an announcement alone. It lies in a stacking of smaller things: release notes in which functionality shifts from "entering" to "checking," pricing models that charge per processed unit instead of per user, support documentation that speaks of approving and rejecting instead of executing, and job postings at the supplier itself that shift toward AI oversight instead of AI building. Each signal on its own is weak. Together they show whether a supplier is moving from category three (human work) to category two (oversight) or category one (full takeover).
Suppliers that directly touch your largest cost items or work for which you have capacity of your own deserve a fixed viewing frequency: every quarter is a workable rhythm for most sectors, because product updates and price changes usually do not occur more often than that. For suppliers further removed from your core process, a semi-annual look is sufficient. What does not work is looking once during procurement and never again, because today's product is not the product with which you signed the contract.
If a supplier takes over, within the product itself, a task that you currently have placed with a team or an external party, that means the assumption "this work requires people, from us or from a third party" has lapsed for that specific piece of work. It does not immediately mean that the capacity currently doing that work has become redundant: oversight is often still needed, and the transition happens in parts. It does mean that the budget line tied to that work needs to be re-examined for what it actually still covers.
An AI feature in a supplier's software is not proof that the work in your organization changes in the same way. It depends on how your process is structured, which exceptions you allow, and how much oversight your sector or regulation requires. A supplier that claims a task is fully taken over by AI is describing what the product can do under ideal circumstances, not what happens in your specific setup. That distinction — what the supplier promises versus what actually happens at your organization — is precisely where the mistake usually arises.
The signal itself does not lead to a personnel decision. Where that is at issue, its own statutory requirements and its own process apply; that is a matter for the employer, not for this signal. What the signal does feed is a strategic decision: do you continue to independently fill a task that a supplier now delivers as standard, or does your team's role shift from executing to checking. That decision does not belong to a single officer; who should take part in a strategy review partly determines whether this signal reaches the boardroom table in time, rather than only at the annual accounts.
What a supplier builds into its product is connected to what your own people are already doing with the same technology: sometimes employees are ahead of the supplier's offering and use separate AI tools alongside the existing system, which can be seen via how do you track what your own people are already doing with AI. It is also connected to price: a supplier that builds in AI often also changes its revenue model, and you track that separately via how do you track the price of AI capacity. And it is not only relevant among existing suppliers: new parties that offer a comparable service without building up personnel deserve a view of their own via how do you track new entrants without personnel.
All these signals ultimately do not answer the question of which work within your own company can genuinely be taken over by AI; that is a different question, and it is answered per task with the work scan from FTE TO AI.
You can start with a list of the suppliers that directly touch your largest work items, and note per supplier when you last checked what their product now does on its own. If you want to know which assumptions underlying your own strategy are most sensitive to signals like these, you can request a free assumption check: a short conversation in which you name your key assumptions and see, per assumption, when it was last confirmed. The full proof print, with an ongoing view alongside your management team's self-plot, is under construction.
Stel uw vraag. Vaak zit de echte vraag een laag dieper — daar mag ik naar vragen.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.