AI for professional services: the groundwork done, the judgement yours

AI takes over the work around the advice: reading and organising documents, summarising and comparing paperwork, finding knowledge from earlier files, handling intake and preparing reports. That applies to accountancy firms, law firms and consultancies. The judgement, the responsibility and the client contact stay with the professional; AI does the groundwork.

Most of the time saved is in the work before and after the substantive review: requesting documents from clients and checking them for completeness, processing bank statements and invoices, flagging anomalies and preparing recurring reports. An AI agent does that groundwork and the accountant checks. The closest work we have done is EVOKE, a finance platform for SMEs built together with Ficare: it connects to the Dutch ERP system AFAS, runs checks automatically, flags anomalies and answers questions about the figures with the reasoning behind them.

Legal AI, also called legal tech with AI, is software that reads, summarises and compares legal texts. For law firms it helps with document analysis, with a first check of contracts for deviating or missing provisions, with finding earlier documents and advice in your own files and with preparing draft texts. The lawyer assesses, amends and signs. AI does not give a legal judgement.

AI can do the groundwork for a contract review: read the contract, extract the key provisions, compare them with your standard clauses or a checklist and mark deviations and missing elements. A lawyer assesses those points and decides. The same principle is running at du Gardijn Verzekeringen: an AI agent compares old and new policy schedules, shows the differences and prepares a draft email that a member of staff checks.

Confidential files do not belong in a public chatbot. We work with business AI models that do not use your data for training, choose processing within the EU wherever possible and record the processing in a data processing agreement. Access follows the permissions that already apply: anyone who may not open a file gets no answer from it through AI either. Whether an application fits the professional rules of your field is something you assess in advance; for a legal review, work with a lawyer.

The professional, and that is why we build it in that way. For anything with legal, tax or financial weight, human oversight (human-in-the-loop) is the starting point: AI delivers a draft, a summary or a flag with the source included, and an accountant, lawyer or adviser checks and approves before anything goes to a client. That way you can see what an answer is based on, and a mistake is not adopted unnoticed.

Anyone deploying AI takes measures to support the AI literacy of their people. That obligation under Article 4 has applied since 2 February 2025, and since 27 July 2026 in the lighter text of the Digital Omnibus on AI (Regulation (EU) 2026/1744). Since 2 August 2026 it must also be recognisable when clients are talking to your AI or seeing realistic AI-generated images. You are usually the deployer, not the provider, and most applications fall into minimal or limited risk. The heavy requirements for high-risk AI, such as AI in recruitment or credit scoring, apply from 2 December 2027. Innoworks is an implementation partner and does not give legal advice; for a legal review, work with a lawyer.

There are no cases from accountancy firms or law firms on our site yet. Our evidence comes from adjacent professional services: mortgage and insurance advice (HypotheekCompany, Bedrijfshypotheek.nl and du Gardijn Verzekeringen), finance for SMEs (EVOKE, built together with Ficare) and strategy and marketing agencies (LAB63 and DigiSocial). The work there is the same as in accountancy and legal practice: reading documents, unlocking knowledge from your own files and preparing reports, with a professional who checks. Those cases are on this page.

That is the starting point. We do not replace your systems; we build applications and integrations that connect to them, so the output lands where your team already works. Which integrations are feasible for you is something we assess in advance, because not every package offers an integration.

With one process that has a clear owner and measurable time spent, such as getting files complete or finding earlier advice. In a fourteen-day AI audit we map your processes and deliver a prioritised opportunity report with the expected return per opportunity, assessed for privacy and feasibility. A first, clearly defined AI agent is usually live within two to four weeks.

That depends on the process, the number of files and the systems it connects to. We agree a fixed scope and a fixed price up front, so you know where you stand. We often start with an AI audit that ranks the opportunities by return. Book an intro call to talk through your situation.