AI and custom work: what changes in the assessment of your work

Custom work is not an exemption

Many directors assume that custom work automatically means that AI can do little with it. That is partly true, and partly not. Custom work is not a characteristic of an entire company, it is a characteristic of individual tasks. And within those tasks there is often a structure that does repeat, even when the outcome looks different each time.

A custom quote consists of a client request that is always different, and of calculation rules, price tables and delivery times that are not. A technical advisory consists of a unique situation on location, and of a fixed assessment framework against which that situation is tested. The custom work lies in the first part. The second part is exactly the part that can be taken over, entirely or with oversight.

Why two companies with the same tasks come out differently

Two companies may have the same tasks on paper — drawing up quotes, answering client questions, providing technical advice — and still get a completely different outcome in a work scan. That difference does not arise because one is 'more custom' than the other. It arises at three points:

How fixed the assessment framework is. If a company works with fixed standards, tables or protocols that underpin the custom decision, that framework can be transferred to a system. If that knowledge only exists in the heads of two senior employees, there is nothing to take over until it has been documented.
How repeated the input is, despite variation in the output. A hundred quotes that all end differently, but all start with the same ten questions, produce a different picture than ten quotes a year that are each built up from scratch.
What the cost of errors is if something goes wrong. Custom work in an environment with low error costs leaves more room for a system that produces a first draft. Custom work with high error costs — an incorrect sizing, a missed standard — more often requires assessment by a human before anything goes out the door, as shown with work with high error costs.

The work scan by FTE TO AI breaks down a company's work into tasks and assesses each task on eight axes. With custom work, it is specifically the axis of repetition and the axis of error costs that determine the outcome — not the label "custom work" itself.

Which tasks shift, and which do not

Within custom-work companies we consistently see the same pattern recur, regardless of the sector.

Shifting to AI, entirely or largely:

writing out the intake conversation into a structured request specification
price calculations based on fixed tables and current purchase prices
a first draft quote or advisory, based on comparable previous cases
aligning planning and availability against existing capacity
reporting status and progress back to the client

Shifting to the middle ground, where AI makes a proposal and a human assesses it with reason:

the translation of a client's wish into a technical specification
deviations from the standard setup, where the assessment framework leaves room for interpretation
correspondence in which the client asks something that falls outside the usual pattern

What remains human work:

the first contact in which a client does not yet know exactly what he wants, as discussed with work with a lot of client contact
the final responsible assessment in borderline cases, especially in work that falls under supervision
the execution itself, if it is physical or negotiating in nature

This pattern is not universal. In the installation sector, for example, the outcome depends heavily on how standardized the calculation already is, something discussed more extensively at what AI can take over in the installation sector. For the administration of a custom-work company the ratio is often different again, because invoicing and file building are generally less client-specific than the primary process itself, see what work can AI take over in administration.

What this is not

These outcomes say something about the task, not about the employee who currently performs the task. Whether and how an organization does something with hours freed up is a choice for the employer. If that choice touches on positions or on personnel, the employer's own legal requirements apply; this page and the work scan do not address that. We provide an indication of work, not a basis for a decision about people.

It is also worth noting where the boundary of this assessment lies. Custom-work companies differ greatly in how consistently their own framework has already been documented. A company where the assessment framework only exists implicitly, in the experience of a few people, will see tasks shift less quickly than a company where that framework is already on paper. That difference cannot be fully captured with a short questionnaire, but it can provide an initial direction.

What you can do now

The free quickscan consists of twelve questions, no account required, and gives an indication of what share of the hours in your profile can be taken over by AI today. For a custom-work company, that indication is a starting point: it shows whether your work, in terms of repetition and error costs, leans more toward fixed frameworks, or more toward ever-new judgment calls.