In professional services, the work largely consists of language: drafting documents, reviewing files, formulating advice, tailoring reports to a client or situation. Law firms, accounting firms, consultancies and insurance intermediaries differ in profession, but not in this pattern: many hours go into gathering and rearranging information that already exists somewhere, and a smaller part into the judgment that follows. That makes the sector sensitive to a shift that is already underway: language work is exactly the kind of work AI systems score well on, provided the information they work with is accessible and reliable.
The circumstances that determine the outcome do not lie in the profession itself but in how the firm is organized. If a team works with fixed templates and a file structure that everyone uses in the same way, a large part of the work can be made predictable. If the team works bespoke per client or per case, with deviations that exist only in the head of the handling professional, that work is harder to hand over, regardless of how ambitious the objective is.
Within the same sector, firms already differ widely. At one firm, a system drafts an initial draft opinion, with an employee reviewing it and approving or returning it with reasons. At another firm, exactly the same work is still done entirely by hand, not out of unwillingness but because the files are not in a form a system can read, or because no one has established when a draft opinion is good enough to be passed on.
Three categories of work run through every firm. Part of it a system can take over independently: summarizing documents, filling in standard forms, flagging anomalies in figures. Part of it can be done partly, with an employee reviewing the outcome and approving or rejecting it with reasons. And part remains human work: the conversation with the client about a difficult situation, the judgment call that cannot be captured in rules, the responsibility that comes with a name and a signature. Which task falls into which category differs per firm and per process, and that is precisely why an ambition such as "we are deploying AI for our advisory process" says little without further explanation.
A sentence such as "AI will speed up our file work" sounds unambiguous, but in the boardroom it is rarely heard that way. One partner thinks of a system that supplies draft texts that are still rewritten in full. Another thinks of a process in which the system works largely independently and is only referred for review in case of doubt. Both readings are a reasonable interpretation of the same ambition, and the difference between them determines how much capacity is freed up, which roles actually change, and which decision rights go with that. How exactly that difference arises, and why it so often goes unnoticed, is described on the page about why people mean different things by the same words.
This ambiguity also affects the question of how the organization itself is structured: which department will do the work going forward, which layer manages it, and where which decision is made. Anyone who wants a firm picture of this before the ambition is set down will find an explanation without jargon on the page about a target operating model in plain words.
An ambition that assumes AI work asks different things of an organization than the same ambition without AI. For professional services, that comes down to, among other things: are files and templates in a form a system can read, is there an established moment at which an employee reviews an AI outcome and approves or rejects it with reasons, and is it clear who remains responsible when the final advice goes out the door. On these points it often turns out that the ambition moves faster than the organization: management speaks of automating the advisory process, while the underlying files are not yet structured enough to be read by a system. That is not a matter of willingness, but of readiness, and that readiness can be tested per dimension separately.
Other sectors show a comparable pattern with their own circumstances: in education different ambitions play a role than here, as described on the page about AI ambitions in education, and in retail too the work shifts along a different kind of lines, set out on the page about AI ambitions in retail. The underlying mechanism — which work can be handed over and which work remains human work — is the same in every sector, only the specifics differ.
This page describes which work can shift, not what a firm should do with its staff if that happens. Decisions affecting the employment relationship are subject to their own legal requirements, independent of which task a system can or cannot take over.
The question of which work at this specific firm can actually be handed over to AI cannot be answered with a general impression of the sector. The work scan from FTE TO AI answers that question per task, with an outcome that indicates which part of the work is done independently, under supervision, or still by people. Anyone who wants to know first where their own organization stands furthest along and where it lags behind can take the free readiness check: eight short questions, one per dimension, giving a picture of where readiness already stands and where a further step is needed. The full ambition assessment, with the four layers from vision to target state and the translation into roles and decision rights, 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.