"We need to deliver faster." In a management meeting, that sounds like an unambiguous ambition. In practice, one person means: shorter lead times with the same headcount. Another means: the same volume with less capacity. A third means: responding faster to requests that currently sit unanswered. All three are legitimate, but they ask different things of the organization.
The difference becomes sharper once AI enters the picture. "Delivering faster" without AI usually means: simplifying processes, removing waiting times, scheduling people differently. "Delivering faster" with AI work built in means: certain steps are no longer done by a person, but by a system, with or without someone reviewing the result. That is a different ambition, even if the words are the same.
When accelerating delivery times, three situations run together, often within the same process:
Companies where this already works have unraveled these three categories down to the level of the task. Companies where it does not yet work still talk about "the process" as a whole, and that is exactly where the delay remains: you cannot speed up a part if you do not know which part it is.
Delivering faster with AI work built in does not primarily require more technology. It requires eight things that are interconnected, and that are usually unevenly developed within an organization:
1. Task decomposition — the ability to break down a process into steps that can be assessed individually for AI suitability. 2. Data readiness — the input a system needs is not always the input that is currently available or structured. 3. Oversight design — who approves or rejects, on what grounds, and what happens with a rejection. 4. Decision rights — if a system makes a proposal, it must be established who has the mandate to confirm it. 5. Error handling — delivering faster without a working recovery path for errors only produces faster errors. 6. Measurability — without a baseline measurement of current lead times, an improvement cannot be demonstrated, only felt. 7. Change capacity — people who currently perform the full task need to learn to assess the output of a system; that is a different skill. 8. Customer expectation — delivering faster changes what a customer subsequently expects, and that has knock-on effects in other parts of the organization.
Together, these eight dimensions determine whether "delivering faster" is a realistic ambition for the coming year, or a wish waiting for capacity that does not yet exist.
The ambition changes who decides about it. Alongside the process owner who currently monitors lead time, there is a need for someone who determines and keeps up to date, per task, the boundary between "AI can do it", "AI makes a proposal", and "remains human work" — that boundary shifts, and not automatically in the same direction. A role is also needed for someone who assesses the quality of AI proposals over time, separate from whoever is providing oversight on a given day. Without making that role explicit, oversight remains dependent on individual attentiveness, and that is not something on which management wants to base a speed commitment.
This does not touch on personnel decisions as such; whatever consequences an organization attaches to this falls under its own applicable legal requirements.
No percentage tells the whole story here, and exactly how much time is freed up depends on the process, the data quality, and the degree of oversight the organization finds acceptable. What can be recognized, however: a recurring process in which the number of hours freed up per team is made visible, a fixed point where rejected AI proposals are discussed rather than ignored, and customers experiencing a shorter lead time without a noticeable increase in the error margin. If those three things are visible, the ambition has not only been expressed, but also realized.
Delivering faster rarely stands alone. The same eight dimensions recur, with different emphases, in expanding internationally and the capabilities that requires, in integrating an acquisition without confusing the readiness of both organizations, and in becoming more customer-oriented when AI work plays a role in it. Anyone who notices that their own vision is still too vague to apply this check to it will find pointers in making the vision concrete before you test it.
The underlying question — which work in this company can genuinely be taken over by AI — is answered by FTE TO AI's work scan per task, not per process or department. If you first want to know where your organization stands on these eight dimensions, there is the free readiness check: eight short questions, one per dimension, resulting in a picture of where you are furthest along and where you are least far along. The full ambition check, with the four layers and the five confidence gates, is under construction.
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.