Knowledge base

Implementing AI at a notarial office

Implementing AI at a notarial office need not be an IT project lasting months. The three conditions are a clear AI policy, a secure environment in which privacy-sensitive data never leaves your own environment, and a phased start alongside the existing systems. Offices that begin with one clearly defined workstream, measure the results and then broaden are typically operational from day one: no data migration is needed and roll-out runs via Microsoft SSO.

Why is implementing AI not a months-long IT project?

Many notarial offices postpone AI because implementation sounds like a heavy IT undertaking: a project team, a migration, months of lead time. That assumption is wrong. Modern AI software for the notarial profession runs as an operational layer alongside your existing systems and requires no rebuild of your IT landscape.

What the implementation does require are three conditions. First, a clear AI policy, so that everyone at the office knows what is and is not allowed. Second, a secure environment: privacy-sensitive data never leaves your own environment and the Data Shield anonymises personal data before processing, in line with the duty of confidentiality under Article 22 of the Wna. aiNotaris itself is certified to ISO/IEC 27001:2022.

The third condition is a phased start: first one workstream, then measure, then broaden. This article sets out that approach: why your IT chain sets the pace, why a cloud migration is not yet an AI foundation, and which phases a successful implementation goes through.

Why does your IT chain set the pace?

Anyone implementing AI deals with two fixed partners: the vendor of the NSL/DMS and the external IT administrator, the MSP. The white paper IT-Landschap Notariaat 2026 analyses their role soberly: both act on entirely valid business interests, and precisely those interests limit the office's pace of innovation.

Software vendors generally do have APIs, but weigh strategically how far to open them to independent AI platforms; they prefer to offer an all-in-one environment. For the notarial office this means, as the white paper puts it, that the pace of innovation is directly capped by the development capacity and the willingness of that one vendor.

IT administrators carry the responsibility for continuity, security and compliance, with business models historically woven into local infrastructure: servers, firewalls, workplace security. Cloud-native AI that runs in the browser reduces the need for that infrastructure. This explains why IT administrators can be hesitant about independent AI systems and prefer to steer towards the vendors' familiar roadmap.

For your approach this means: do not wait for the chain to take the initiative, but take charge yourself. Choose a solution that works alongside the existing systems, so nothing has to be replaced in order to start, and involve your IT administrator early with concrete questions about security, management and access. A good vendor of the AI layer supports that conversation and supplies the technical documentation your administrator needs to feel at ease.

Is a cloud migration the same as an AI foundation?

No. Many offices see the cloud migration of their notarial package as the logical first step towards AI, but the white paper adds an important caveat: the wider business world set its move to the cloud in motion back in 2008. The belated migration of notarial software is therefore primarily the closing of a historical IT gap, not fundamental innovation.

More important still for your implementation: a migration does not change the nature of the software. The system remains reactive and the database structure predominantly linear. For proactive AI that orchestrates processes independently to perform at its best, an AI-first architecture is needed: a system designed around AI from the ground up. The term comes from Gartner's article AI-first, and the white paper applies it to the notarial profession.

The practical consequence is reassuring: you do not have to wait for your vendor's cloud roadmap. An independent AI layer with its own AI-first foundation runs alongside your current environment today, whether it sits on-premise, runs in Citrix or has already moved to the cloud.

What does the step-by-step plan look like?

A successful implementation runs in four phases. Each phase has a clear result, and none of them requires a migration or the replacement of existing systems.

Phase 1: policy and AI literacy. Set down what staff may and may not do with AI and train the team in the basics. Article 4 of the EU AI Act (Regulation (EU) 2024/1689) has required organisations to ensure sufficient AI literacy since 2 February 2025. The KNB AI-Weegschaal and the free model AI protocol from aiNotaris provide a directly usable framework.
Phase 2: choose one clearly defined workstream. Start where the volume is high and the result measurable, for instance inbox management, document review or meeting preparation. One workstream keeps the change manageable and quickly gives an honest picture of the time saved.
Phase 3: work alongside your NSL/DMS, without migration. Your file management system remains the source. Roll-out is plug-and-play via Microsoft SSO, without complex server installations, and no lengthy data migration is needed: as soon as the accounts are linked, the team can get going.
Phase 4: measure and broaden. Before you start, record how much time the chosen workstream costs today, measure the difference after four to six weeks and only then expand to the next processes. That way usage grows on the basis of results, not enthusiasm alone.

Offices that keep to this order are operational from day one and then broaden at a pace the team can handle. If you prefer to have the frameworks in place first, start with phase 1 today: the policy and the training can run entirely in parallel with the selection of the software.

How do you bring the team along?

Technology is rarely the reason an AI implementation stalls; adoption is. So train the whole team, not just the front runners. The learning curve is deliberately kept low, but a joint start prevents usage from depending on a few enthusiastic colleagues.

Also appoint ownership. For each workstream, designate one member of staff who tracks usage, collects questions and shares successes with the rest of the office. Without an owner every change fades, however good the software.

Finally, keep the human-in-the-loop principle visible in everything you communicate: the software proposes, the notary reviews and signs. That is the line of the KNB AI-Weegschaal, which takes the preservation of human oversight as its starting point, and it is also the message that reassures hesitant staff. The digital colleagues take over the preparatory work; the judgement remains human work.

Frequently asked questions about AI implementation

No. The cloud migration of your notarial package is a modernisation step, not a precondition for AI. An independent AI layer such as aiNotaris runs alongside your current environment, even if it sits on-premise or in Citrix.

That functionality may become a fine addition, but your pace of innovation is then tied to the development capacity and priorities of a single vendor. An independent AI layer lets you start today, without saying goodbye to your trusted package.

Choose a process with high volume and a measurable result, such as inbox management, document review or meeting preparation. One clearly defined workstream gives an honest picture of the time saved within a few weeks and keeps the change manageable for the team.

Article 4 of the AI Act has required organisations, since 2 February 2025, to ensure sufficient AI literacy among staff who work with AI. An adopted AI policy and a basic training session for the team give that practical substance; the free model AI protocol is a useful starting point.