AI in professional practices: it's already in your software (and you're not using it)

For a professional practice — an accountant, a labour consultant, an associated firm — the AI question is different from almost every other sector. It isn't "which tool do I buy": it's "why am I not using the one I already have". Because, nearly always, AI is already inside the practice-management software you pay for. The vendors that dominate the market — TeamSystem, Zucchetti, Wolters Kluwer with Bluenext, DataLog — all shipped, between 2025 and 2026, native AI features inside their software. This isn't a market where AI has to be built from scratch: it's a market where AI arrived on subscription, and largely sits unused.

The data confirms it: 34.1% of Italian professional practices already use AI tools consistently, and the projection is to reach 71.9% within three years. This isn't a niche fad — it's half the sector on the move. And it doesn't compete with other investments: practices' ICT spend reached 2.01 billion euro in 2025 (+3% year on year, with a further +4% projected in 2026). AI is riding a digitisation budget that is already growing, not opening a new one. The right reading, for a practice, is counterintuitive next to any sales demo: the missing piece isn't the tool, it's the workflow around the tool.

AI is already in the software you pay for (the problem is elsewhere)

Almost every article on AI starts with "here's what you should buy". For a practice, that's already the first mistake. The vendors a firm uses every day have integrated AI into the product: automatic reading and categorisation of accounting documents, deadline monitoring, document triage. These are bundled features — included, not sold separately. The problem, therefore, isn't access: it's that no one has switched them on, configured them and slotted them into the way the practice actually works.

It's the exact opposite of manufacturing, where the brake is cost and the State co-funds adoption. Here the cost is already borne — it's inside the software subscription. What's missing is the process redesign: who uploads what, where AI reads and proposes, where the human steps in, how the saved hours flow back into the practice's work. It's the theme of every operational function: the tool is there, the return depends on the design of the flow. And it's precisely what we mean by "we implement it".

Where it frees up hours, for real

Three flows, in a practice, are high-volume, repetitive and rule-based — AI's natural terrain, and almost always already covered by software features you're paying for:

• Reading and categorising invoices and receipts (OCR) — instead of manually entering data into the software. The native AI modules read, categorise and import documents, cutting hours of data entry per client. It's the flow with the most immediate, most visible return.
• Fiscal-deadline monitoring — instead of a tickler kept by hand, client by client. Automated systems track every deadline, send graduated reminders and chase documents by email, without anyone having to hold it all in their head.
• Document and compliance triage — instead of manually sorting the paperwork clients send in. It's the same AI layer of the software, included and not extra, that orders and routes what arrives.

The criterion for starting isn't "which is the most impressive", but "which do I already have that takes away the most repetitive hours": for most practices the answer is accounting OCR — high volume, clear rules, a before/after measurable in hours given back.

The real value: freed hours = more advisory

Here is the part that changes the conversation, and that almost no vendor tells the right way. A practice doesn't buy AI to let someone go or to cut costs: it buys it — when it buys well — to sell more advisory. The time freed from data entry and sorting doesn't vanish from the P&L: it shifts to the work the client happily pays for and that no software replaces — strategic advisory, planning, the relationship.

It's the reason why, in a practice, "AI adoption" and "margin growth" are the same sentence said twice. Every hour taken away from data entry is an hour that can become billable advisory at higher value. Whoever reads AI only as saving grasps half of it; whoever reads it as advisory capacity recovered grasps the point — and it's the reading on which we build the design of the flow.

The numbers, read honestly

The picture is one of a sector on the move, not a gamble. 34.1% of practices already use AI consistently, on course for 71.9% within three years: whoever waits doesn't avoid the change, they postpone it while half of their peers put it into practice. And the investment isn't a leap into the void — it's inside a budget that grows on its own: 2.01 billion euro of practices' ICT spend in 2025, +3% year on year and a further +4% expected in 2026.

The honesty, though, is in not confusing spread with return. That half the sector "uses AI" doesn't mean half is making the most of it: it means the features are there, often half switched-on and with no flow around them. The gap, as in every department and every sector in this series, isn't between those who have AI and those who don't — it's between those who have redesigned one process around a feature they already own and those who leave the tool running empty. The distance between the 34% who "use it" and the practice that truly gains is all there.

And compliance? Within limits, but with method

A practice handles clients' fiscal and personal data: compliance isn't a detail, it's the job. The transparency, documentation and human-oversight duties set out by the EU AI Act for limited- and high-risk systems apply directly to the use of AI in a professional practice — it isn't a free zone just because the feature is included in the software. Clarity on where the vendor processes and stores the data, minimisation, and above all a documented human oversight — the professional responsibility stays with the professional, AI reads and proposes, the signature and the decision stay with whoever answers to the client and the tax authority.

It's exactly what our compliance overlay wires to every workflow we design: so that the freed hours come from a flow that is not only efficient, but also defensible. For a practice, this isn't extra bureaucracy — it's the same care it sells to clients, applied to its own tool.

Where to start, in practice

If you run a practice and AI is in your sights, the sensible path is short and — surprise — almost always at a cost already borne:

• Start from what you already pay for — map the native AI features of your software before considering any new purchase. The first value is switching on and configuring what's included, not buying more.
• Choose a single high-volume flow — accounting OCR or deadline monitoring give the sharpest before/after in hours. Not a rollout, one flow.
• Measure the hours, not the features — the number that counts is how many data-entry hours come back available per client, not how many checkboxes you've enabled.
• Reinvest into advisory — the hours recovered become billable advisory at higher value. That's where AI, in a practice, makes margin.

Before even choosing the flow, though, it's worth knowing where you stand: our AI-readiness assessment helps work out which feature to start from for the most return and least friction, and which controls to put around the first flow. If the topic is the compliance of what touches client data, our compliance overlay explains how we wire the controls to every design.

We've turned the first step into a free, self-serve assessment: a few questions and a pointer on where to start, with which controls around it. Take the AI-readiness assessment — then, if it makes sense, let's talk.

This article is for guidance only. The adoption and spend figures cited (34.1% of practices using AI consistently, projected to 71.9% within three years; 2.01 billion euro of practices' ICT spend in 2025, +3% year on year and +4% expected in 2026) come from sector sources and should be read as indications of direction, not as guarantees of results. The AI features of individual software packages, their availability in your own plan and the applicable compliance duties must be verified against the vendor's contract and the official texts before any choice. Any automation touching clients' fiscal or personal data must be assessed on the data, the controls and the context of the individual practice.