The ICT sector consists of work that can easily be broken down into repeatable steps, and work that resists exactly that. Writing code, testing, documenting, triaging tickets, reviewing monitoring, rolling out configurations: many of these tasks have a fixed structure and a clear evaluation criterion. Alongside that sits work that revolves around architectural choices, client context, security trade-offs, and the decision of whether or not a system goes into production. That distinction determines where AI is already taking over tasks today, where oversight remains necessary, and where human work remains the core of the matter.
The circumstances that drive the outcome are not uniform. A team that works with a well-documented codebase and clear testing standards can hand over tasks more easily than a team that runs on tribal knowledge and verbal agreements. An organization with a fixed review structure can embed oversight of AI output into an existing process; an organization without that structure must first build it before transfer is responsible.
In parts of the sector, the shift is already visible. Code suggestions, test generation, and documentation updates are, in some teams, largely produced by AI, with a developer approving or rejecting based on functionality and readability. In other teams, the exact same work is still done entirely by hand, not because it is fundamentally different there, but because the conditions are missing: no access to the right tools, no established quality criteria, no time set aside to rebuild the process.
That difference between companies rarely lies in the technology itself. It lies in readiness: does the organization have its data and documentation in order, has a decision-making authority been assigned for who evaluates AI output, has capacity been freed up to set up the new process before the old one is phased out. Where these conditions are missing, an ambition remains on paper, even when the technology is already available.
An ambition such as "AI supports our developers" sounds straightforward, but in practice requires a series of capabilities that are not automatically present: a test environment that can validate AI-generated code, a review process that does not slow down due to extra volume, and evaluators who know on what grounds to reject an AI proposal. Without these capabilities, the ambition remains an intention without execution.
That is precisely why the same words can mean different things within a management team. The CTO thinks of automated pipelines, the operations director thinks of fewer freed-up hours for support, the HR manager thinks of a different role composition. None of these interpretations is incorrect, but without a shared benchmark, the ambition remains noncommittal. How to translate a vision into something measurable, rather than into three separate assumptions, is described in how to make a vision measurable.
The ICT sector is not unique in grappling with this shift, although the bottlenecks differ. In financial services, the question plays out differently, because oversight and accountability play a larger role there; what applies there is described in which AI ambitions are at play in financial services. In construction, physical work is a fixed factor that limits AI takeover differently than in ICT, as described in which AI ambitions are at play in construction. And in the cleaning industry, it is mainly about planning and quality control rather than code production, worked out in which AI ambitions are at play in the cleaning industry. The pattern is always the same: the ambition only becomes concrete once it is clear which capabilities are missing.
A management team that decides AI should take over "most of the review cycle" is thereby formulating an ambition that cannot be realized within a quarter. That is not an objection to the ambition itself, but a signal to phase it: which capability needs to be in place first, which one after that, and at what point is the organization first able to assess whether it is working. What such phasing can look like is described in how to phase an ambition that is too large for a year.
We do not weigh in on personnel decisions that may follow from such phasing: which work an employer discontinues or redistributes falls under its own legal requirements and its own judgment. What we describe is which work can be transferred, partly or fully, and under what conditions.
The underlying question is not whether AI ambitions in the ICT sector are realistic, but whether this specific company, with this codebase, this team, and these processes, is ready to carry them out. Which work in this company can actually be taken over by AI, task by task, is answered with the work scan from FTE TO AI.
Anyone who wants an initial impression first, without obligation, can take the free readiness check: eight short questions, one per dimension, providing a picture of where the organization is furthest along and where it is least far along. The full ambition assessment, with all four layers and the five confidence gates, is under development.
Vertel wat u wilt bereiken, dan kijken we samen wat daarvoor moet staan.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.