fairlane.systems

NOTARIES · INDUSTRY HUB

AI for Swiss notaries: precedent research, civil-code templates and notarisation preparation

How Swiss notaries use AI for case-law research, notarisation preparation and Civil Code estate templates – with anonymisation duty and without delegating the notarisation act itself.

Researched & fact-checked by: · As of: 2026-05

Notarial practice in Switzerland and AI: a cantonally fragmented picture

The Swiss notarial profession is organised cantonally (Art. 55 Title-1 CC) and varies significantly. Three models exist. First: the free-profession (Latin) notariate, as in Geneva, Vaud, Ticino, Bern or Fribourg. Second: the state notariate, as in Zurich, Basel-City or Schaffhausen, where notaries are state authorities. Third: mixed models, as in Aargau or St. Gallen. Switzerland counts an estimated 1,500 to 2,000 licensed notaries; the Swiss Federation of Notaries (SNV) is the central body.

AI in notarial practice in 2026 has its own characteristic. Notarisation itself – the actual notarial act – is a trust-reserve and not delegable. The notary vouches for the identity of the parties, their capacity, free will and the legal compliance of the act. No AI in 2026 can or may assume these tasks. Where AI does contribute usefully: preparation of notarisations, case-law research, estate templates, pre-checks on land and commercial registers, client query triage.

The sector in 2026 is more reserved than the bar. The SNV has published initial statements on AI use in 2025; individual notarial firms, especially larger ones in French-speaking Switzerland, deploy AI productively. State notarial offices remain more cautious due to IT procurement rules. A binding SNV guidance at the depth of the SAV bar regulation is missing in 2026 – the sector aligns with cantonal notarial regulations and with the Civil Code's requirement of notarial authentication.

Why AI is worthwhile for notaries in 2026 nonetheless

Three realities hit notarial practice simultaneously.

First: heavy preparation effort per notarisation. A marriage contract, an inheritance contract, a deed of gift, a real-estate transfer – all require careful preparation with extracts, templates, legal review and correspondence. The notarisation itself is only the tip. AI-supported preparation can save several hours per case and free the notary for more appointments.

Second: precedent and policy knowledge is distributed across the office. The case-law collection, cantonal notarial regulations, Civil Code estate templates, tax aspects of real-estate transfers – knowledge sits with senior notaries. A RAG knowledge base makes it accessible to all staff. Important: the evaluation stays with the notary.

Third: client expectations move. Clients in 2026 expect initial advice by email in two days instead of two weeks, transparent status communication and online rescheduling. A notarial office that does not deliver loses SME mandates (company formation, statute changes) and high-net-worth private mandates (estate planning, foundation formation) to digitally more advanced offices.

Fourth: anonymisation duty is non-negotiable. Notarial files contain especially sensitive personal data (family circumstances, wealth, health questions for advance directives). A prompt to a US-hosted model with unanonymised data is a data-protection incident with potential notarial disciplinary measures. Pseudonymisation or EU/CH hosting with DPA is the 2026 standard.

The point: AI in notarial practice is a tool for preparation and research, not for notarisation. In that role it is mature for productive use in 2026.

Where AI works productively in a Swiss notarial office in 2026

Five application clusters cover the bulk of realistically automatable preparation work.

Case-law research with RAG. The Federal Supreme Court case-law collection (1875 to present) and cantonal decisions are indexed in a vector database. Questions like "Which Federal Court decisions exist on disposition restrictions in inheritance contracts" return grounded answers with citations. The case-law collection itself is public and contains no identifiable client data – a relatively low-risk pilot use case. See Retrieval-Augmented-Generation.

Notarisation preparation. From the parties' answers (online form, email, initial meeting) a draft of the notarial act is generated: marriage contract, inheritance contract, last will, deed of gift, real-estate transfer. Important: AI generates the draft, the notary reviews, refines and signs. The notarisation itself (reading aloud, signature, vouching) stays unchanged with the human.

Civil Code estate templates with context adjustment. Standard templates for wills, inheritance contracts, advance directives and patient directives are contextualised with the client's family and asset circumstances. The agent proposes clauses (prior heirs, successor heirs, forced-heirship provisions, spousal rights); the notary decides on inclusion. Precondition: pseudonymisation of family data before the model call.

Land and commercial register pre-checks. Before a real-estate transfer or a statute change, an agent extracts relevant points from the current register extract (ownership, easements, liens, board composition, capital structure) and presents them structured to the notary. Saves preparation time per case.

Client query triage and initial-advice preparation. Incoming emails are classified (estate initial advice, notarisation appointment, statute amendment, advance directive), summarised and provided with a draft reply. The notary decides on dispatch and tone. For sensitive initial queries (inheritance dispute, divorce with asset division), the system recognises the category and routes directly to the notary without a draft.

Across all applications: anonymisation or pseudonymisation of client data before the model call, EU/CH hosting with DPA and no-training guarantee, audit-grade trail per Art. 957a CO for the notarial bookkeeping and per cantonal retention duties for the files.

How a notarial office starts with AI – in 6 steps

  1. 01Clarify the cantonal frame: notarial regulation, IT security policy (state notaries), retention duties. Output: a one-page memo on what is permitted in the canton and which procurement routes apply.
  2. 02Draft an internal AI guideline: separation between preparation (AI permitted) and notarisation (human mandatory). Anonymisation and pseudonymisation duties, permitted models, ban on shadow AI.
  3. 03Case-law RAG as pilot: indexing of the public collection, access for all staff. A low-risk entry that already shows effects. Four to eight weeks.
  4. 04Decide hosting architecture: at free-profession notaries a multi-LLM gateway with EU/CH hosting and DPA; at state notaries alignment with the cantonal IT office, possibly local hosting (Llama 3.x, Mistral) on cantonal servers.
  5. 05Preparation workflow as a second use case: notarisation preparation for a specific act type (advance directive, deed of gift) with a pseudonymisation layer before the model call. Eight to twelve weeks.
  6. 06Client consent clause: include in the initial-advice mandate, inform existing clients at the next notarisation. Respect refusal and handle manually-only. Documentation in the file.

Where a notarial office should start in 2026

Three stages, in this order.

Stage 0 – Clarify the cantonal frame. What does the cantonal notarial regulation require on data processing, client files and IT procurement? State notaries are additionally subject to cantonal IT security policy. Free-profession notaries operate more flexibly – the frame derives from the cantonal regulation, the revised FADP and the office's own IT policy. A one-page clarification of these points is the precondition for everything else.

Stage 1 – Case-law RAG as pilot. Realistic is indexing the Federal Supreme Court collection and selected cantonal decisions as a first pilot. The collection contains no identifiable client data, risk is low, the benefit to the office immediate. Four to eight weeks implementation. Output: an internal research tool available to all staff.

Stage 2 – Preparation workflows with pseudonymisation. After a successful first use case: notarisation preparation for a specific act type (e.g. advance directive, deed of gift or company formation). A pseudonymisation layer before each model call, clear separation between preparation (AI) and notarisation (notary). Eight to twelve weeks.

Stage 3 – Scaling with a proprietary template library. Only after two successful use cases: contextualised estate templates, client query triage, land-register pre-checks. Precondition: a documented internal AI guideline explicitly fixing what AI may do and what not.

Important for state notaries: cantonal IT procurement often requires tender procedures and security reviews. This shifts the timeline by six to twelve months. Early alignment with the cantonal IT office is mandatory.

Where AI does not belong in a notarial office in 2026

Three areas where reservation in 2026 is not "conservative" but legally and ethically required.

The notarisation act itself. Reading the deed aloud, vouching for identity, capacity and free will, the signature – these acts are a trust-reserve of the notary (Art. 55 Title-1 CC, Art. 9 CO evidentiary force of public deeds). No voice AI, no avatar, no model can or may replace this. That line will not blur in the coming years.

Confidential advisory conversations with voice recording. Inheritance advice, family conflicts, advance-directive discussions live on confidentiality and trust. Automatic voice recording for AI transcription is delicate not only in data-protection terms (revised FADP) but relationally damaging. If transcription is desired: explicit documented consent and deletion of the recording after processing.

Identity and capacity assessment. The notary is obliged to satisfy herself of the parties' identity and to assess their capacity. AI tools for automated identity checks (video identification, biometric procedures) are permitted in some cantons but do not replace the notary's personal judgement. In cases of doubt about capacity (suspicion of dementia, overwhelming pressure), human judgement remains mandatory.

Particularly delicate and not finally settled in 2026: fully automated drafting of inheritance or marriage contracts without notarial pre-advice. Even if technically possible, this falls within the notarisation duty and is not legally permissible.

Trade-offs

STRENGTHS

  • Case-law research takes minutes instead of hours – senior and junior notaries benefit equally
  • Notarisation preparation noticeably faster, the office can deliver more appointments
  • Estate templates available contextualised rather than in the heads of individual senior notaries
  • Client query triage relieves the secretariat, first response to clients within 24-48 hours
  • Land and commercial register pre-checks systematised – fewer overlooked points

WEAKNESSES

  • The notarisation act stays with the human – AI does not replace a trust-reserve
  • Anonymisation duty forces a pseudonymisation layer before the model call
  • For state notaries, cantonal IT procurement takes six to twelve months
  • No binding sector standard in 2026 – each office must draft its own guideline
  • Client trust can be lost through tone-deaf AI communication – tone remains human

FAQ

May an AI draft a notarisation that I then sign?

The draft, yes. The notarisation act itself – reading, vouching, signature – no. In practice: AI generates the draft, the notary reviews, edits, complements from own judgement and signs in own responsibility. Responsibility for legal compliance and suitability for the specific case stays with the notary. "AI wrote this" is not an acceptable justification in 2026 – neither legally nor reputationally.

Must we anonymise client data before using AI?

For external models: yes. Notarial files contain especially sensitive personal data (family, wealth, health). A prompt to an external cloud provider is disclosure to a third party. 2026 standard: pseudonymisation before model call (names, addresses, social-security numbers are replaced with placeholders; restored after the response) plus EU/CH hosting with DPA and no-training. With local hosting on own servers (Llama 3.x, Mistral) disclosure to a third party falls away; anonymisation discipline remains good practice.

What does the Swiss Federation of Notaries (SNV) say about AI?

The SNV published initial statements on AI use in 2025 and discussed the topic at sector days. A binding guidance at the depth of the SAV bar regulation does not exist in 2026. Sector position as of May 2026: AI as a tool for preparation and research is welcome, the notarisation act itself stays untouched, anonymisation and hosting duties apply. Tracking SNV publications and cantonal notarial supervisory communications is worthwhile.

How do state and free-profession notaries differ in AI deployment?

Significantly. State notaries (Zurich, Basel-City, Schaffhausen etc.) are state authorities subject to cantonal IT procurement and security policy. Pilot projects often require tender procedures, security reviews and cantonal data-protection officer review. Free-profession notaries (Geneva, Vaud, Ticino, Bern, Fribourg etc.) decide autonomously within their cantonal regulation and the revised FADP. Notarisation duties and anonymisation requirements are however identical.

Related topics

LAW FIRM · INDUSTRY HUBAI for Swiss law firms: what works, what does not, and what the SAV guidance saysRAG ON YOUR OWN KNOWLEDGE · SERVICERAG on your own knowledge: answers from your documents – with sources, not made uprevDSG · COMPLIANCErevDSG / revFADP and AI: what the revised Swiss Data Protection Act means for LLM useART. 321 SCC · COMPLIANCEProfessional secrecy (Art. 321 SCC) and AI use: what lawyers, notaries, physicians and auditors must observeCLIENT TRIAGE · USE CASEAI triage for client queries: turning WhatsApp, email and phone into structured casesRAG · AI CONCEPTRetrieval-Augmented Generation (RAG): how AI answers from your own documentsAI-READINESS AUDIT · SERVICEAI-Readiness Audit: where your business stands with AI today – clarified in one to five days

Sources

  1. Schweizerischer Notarenverband (SNV) – Themen und Mitteilungen zur Digitalisierung · 2026-04
  2. Schweizerisches Zivilgesetzbuch (ZGB), Schlusstitel Art. 55: kantonales Notariats-Recht · 2026-01
  3. Bundesgericht – Sammlung der Bundesgerichts-Entscheide (BGE) · 2026-03
  4. Eidg. Datenschutz- und Öffentlichkeitsbeauftragter (EDÖB) – Hinweise zu KI und Datenschutz · 2026-02
  5. Notariat Zürich – Aufgaben und Verfahren des kantonal-staatlichen Notariats · 2026-03

FITS YOUR STACK?

What this looks like in your business – a 30-minute intro call.

Book a call