AI Agents for Lawyers

Practice Areas

AI for Private Client and Estate Lawyers: Wills, Trusts and Family Offices

AI private client lawyers draft from precedent, build asset schedules and explain plans to families; formalities, cross-border succession and tax stay with you.

By Dr. Niklas SchmidtUpdated 13 November 20268 min read
On this page
  1. The trust appeal that was never heard
  2. What AI does well in a private-client practice
  3. Asset schedules and family-office summaries
  4. Formalities and cross-border succession
  5. Confidentiality at the highest sensitivity
  6. Clients who arrive with AI-drafted documents
  7. A setup for a private-client team this quarter

Private client lawyers can use AI safely for three things: drafting wills, trusts and letters of wishes from the firm's own precedents; building asset schedules and family-office summaries from statements and valuations; and turning a complex plan into an explanation a family will read. What stays with the lawyer is every execution formality, every tax consequence and every cross-border question, because those are exactly where the models are weakest and the stakes are highest.

The trust appeal that was never heard

On 24 April 2026 the Alabama Supreme Court dismissed an entire appeal in a family-trust dispute because of the lawyer's "egregious" AI-tainted briefs. The court ordered $17,200 in fees, referred him to the State Bar and barred him from filing anything there without a co-signing lawyer in good standing. In the paragraph where he apologised for the hallucinated authorities, he cited two more cases that do not exist.

It is not an isolated example in this practice area. The record US sanction, $110,204.38 in Couvrette v. Wisnovsky (D. Or., December 2025), arose in a $12 million family-winery elder-abuse suit that was dismissed with prejudice after three briefs contained fabricated citations and invented quotations. The Nebraska Supreme Court suspended an Omaha lawyer in April 2026 whose divorce-appeal brief had 57 of 63 citations defective. Contested trusts, capacity disputes and family fallouts are litigated by private-client lawyers who mostly do not litigate, under time pressure, with the same tools that produced those briefs. The lesson is narrow and absolute: any authority in anything you file goes through a citation-checking protocol, and a general chatbot is never the source of a case.

What AI does well in a private-client practice

Away from the courtroom, the profile is different. Most private-client work is document assembly, extraction and explanation, which are the tasks on which the tools now beat lawyers in benchmarks (Vals found AI ahead on extraction, question-answering and summarisation as early as February 2025, while lawyers still won on redlining).

TaskWorks todayTool typeWhat the lawyer checks
First draft of a will or trust from precedentYesClaude or ChatGPT business tier with your precedents in a Project; Word add-ins such as SpellbookEvery clause against the instruction sheet; formalities; tax
Tailoring standard clauses (powers, protectors, letters of wishes)YesSame, with a clause library as referenceConsistency with the rest of the instrument
Asset schedule from statements, valuations, share registersYesModel with file upload; OCR pass first for scansEvery figure against the source page
Family-office reporting (structure charts, distribution summaries)Yes, for draftingModel with code execution (Claude Cowork, Gemini Notebook)Numbers, entity names, dates
Client explainers of a planYesAny business-tier model with tone instructionsAccuracy and what is left out
Cross-border succession analysisFraming onlyGrounded research tool plus local counselEverything
Execution formalitiesNoNoneLawyer, per jurisdiction
Tax computations and electionsNo, without a tax toolBlue J or equivalent, then a tax adviserEverything

Anthropic's Claude for Legal plugins (12 May 2026) do not include a private-client plugin, so most firms build this on a general model with a Project or Skill holding the precedent bank. Adviser-facing platforms such as Wealth.com now pair document generation with tax-planning modules and custody integrations, which tells you where the wealth managers who refer you work are heading.

Asset schedules and family-office summaries

The most immediately useful workflow is extraction. A family office sends 40 PDFs: bank and brokerage statements, a share register, property valuations, a life-policy schedule, two trust accounts. The lawyer needs one schedule, by entity and by jurisdiction, with dates and sources.

Run scanned documents through a dedicated OCR pass before upload (Claude and ChatGPT read scans by vision, but for numbers a text layer is safer, and Claude in particular tends to say "illegible" rather than invent a figure). Then:

You are preparing an asset schedule for an estate-planning file.
Work only from the uploaded documents. Do not estimate or infer values.

Produce a table with these columns: asset description; legal owner
(exact name as shown); jurisdiction of the asset or account; value;
valuation date; currency; source document and page.

Rules:
- One row per asset or account. Do not aggregate.
- If a value is missing, illegible or out of date by more than
  12 months from today, write "CHECK" in the value column and say why.
- Flag any asset held through a trust, company, nominee or joint
  account in a separate "structure" column.
- After the table, list every question you could not answer from the
  documents, in the order a lawyer would ask them.

Every figure is then checked against the source page before it goes anywhere near a will, a trust accounting or a tax return. The method, including how to handle poor scans, is in extracting data from PDFs. For the tax side, the tax lawyers page covers research tools and the limits of computation.

Formalities and cross-border succession

Two areas in this practice cannot be delegated, and the reasons are structural.

Execution formalities differ by jurisdiction in ways a model will state confidently and wrongly. Whether a will needs two witnesses or three, whether a holographic will is valid, whether a notarial form is required, whether a trust is recognised at all, whether forced-heirship rules override the instrument, which law the EU Succession Regulation applies to a habitual resident who has not made a choice: these are questions with definite answers that vary by country and by year. A short formalities checklist for every instrument, completed by the drafting lawyer, is the control:

  • Which jurisdictions are touched (domicile, habitual residence, nationality, situs of each asset)?
  • Which law governs succession to each asset, and has a valid choice of law been made?
  • What are the execution requirements in each relevant jurisdiction (witnesses, notarisation, language, registration)?
  • Are there forced-heirship or matrimonial-property rules that limit the instrument?
  • Does any jurisdiction refuse to recognise the trust or foundation structure proposed?

The data on AI and multi-jurisdictional questions are not reassuring. In Vals AI's October 2025 legal research study every tool, general and legal, dropped 11–14 points on multi-jurisdictional questions. A 2025 study reported in the AI Law Librarians review found frontier models hallucinating on 45% of Los Angeles questions, 55% of London questions and 61% of Sydney questions, and 100% of the time on one Australian tenancy topic. Use a model to draft the list of questions for local counsel; do not use it to answer them.

Dr. Niklas Schmidt, who teaches the AI Lab for Lawyers course, is the immediate past chair of the IBA Private Client Tax Committee, a member of STEP, and co-editor of an English-language book on crypto taxation covering more than 40 countries, which is one reason the course spends time on how to frame cross-border questions for a model without trusting its answers.

Confidentiality at the highest sensitivity

Private-client files hold what a firm least wants leaked: net worth, family conflict, health and capacity information, beneficial ownership. The controls are the same as elsewhere but the tolerance is zero.

  • No client data in consumer tiers. ChatGPT's Free, Plus and Pro tiers train on conversations unless the user opts out; consumer Gemini keeps activity for 18 months by default and a subset of chats is read by human reviewers, with Google's own advice being not to enter confidential information. Claude's consumer plans make the user choose, with five-year retention if training is enabled. Business, Team, Enterprise and API tiers of all three do not train by default. The full picture is in the confidentiality checklist.
  • Client-side chatbot use is not privileged. In United States v. Heppner (S.D.N.Y., bench ruling 10 February 2026, opinion 17 February 2026) Judge Rakoff held that research memos a defendant generated with consumer Claude were not privileged or work product, because "Claude is not an attorney" and the terms permit logging and disclosure. An English tribunal warned in UK v SSHD [2026] UKUT 81 (IAC) that pasting client letters into public AI tools puts them "in the public domain". Wealthy clients experiment with these tools; part of the engagement conversation is now telling them not to feed the family's affairs into one. See privilege and AI.
  • Meetings. AI note-takers in a capacity assessment or a family meeting raise consent issues; the NYC Bar's Formal Opinion 2025-6 (22 December 2025) requires the consent of all parties and treats not recording as the default. The setup is in AI meeting notes for lawyers.
  • European rules add specifics: the Austrian bar's four practice rules (EU servers, disclose AI use to clients, the lawyer keeps final responsibility, no client data for training) and the German BRAK's position that tools which store client data for training breach § 43a BRAO. Both are in ÖRAK, BRAK and CCBE guidance.

Clients who arrive with AI-drafted documents

The State Bar of Texas's 2026 AI survey found 43% of lawyers reporting that clients raise AI-related topics with some regularity, and 64% naming confidentiality or privilege as their top concern. In private-client work this shows up as a client who has "already drafted" a will with ChatGPT, or a family office that has asked a chatbot to design a holding structure and wants you to "just paper it".

Treat these as instructions, not drafts. Read what the client produced for what it reveals about intentions and assumptions, then start from the firm's precedent. Explain, once, why: the client's document may be invalid on formalities, silent on tax, and built on a hallucinated rule. The CCBE's October 2025 guidance asks lawyers to tell clients when AI is used and let them object; the reverse conversation, about the client's own AI use, is now equally part of the retainer. Guidance on the disclosure side is in telling clients about AI.

A setup for a private-client team this quarter

  • One approved model in a business tier, with the precedent bank loaded as a Project or Skill so drafts start from your clauses, not the internet's.
  • The asset-schedule prompt above, adopted as a team template, with the "check every figure against the page" step written into the file-opening checklist.
  • The formalities checklist attached to every instrument as a matter of course.
  • A one-paragraph client note about the firm's AI use and about the client's own, added to the engagement letter.
  • Training logged for Article 4 of the EU AI Act, which national authorities have supervised since 2 August 2026.

Module 2 of the AI Lab for Lawyers course works through precedent-based drafting and document extraction with this kind of file, using browser tools and no installs. The work that remains for the lawyer after all of this is the work clients were paying for in the first place.

Frequently asked questions

Can AI draft a will or a trust deed?

It can produce a competent first draft when it works from your precedent bank and a structured instruction sheet, and it is useful for tailoring standard clauses and drafting explanatory letters. It cannot be trusted with execution formalities, which differ by jurisdiction, or with tax consequences. Every draft is reviewed clause by clause by the lawyer who signs it off.

Is it safe to put a client's financial information into ChatGPT or Claude?

Only in a business or enterprise tier that does not train on inputs and under a data-processing agreement, and preferably a tool your firm has approved. Consumer tiers of ChatGPT and Gemini train on conversations unless the user opts out, and Gemini's consumer chats can be read by human reviewers. For family-office and trust data, treat consumer tools as off limits.

Are a client's own conversations with an AI about their estate privileged?

Generally not. In United States v. Heppner (S.D.N.Y., February 2026) Judge Rakoff held that a defendant's research memos generated with consumer Claude were neither privileged nor work product, because the AI is not a lawyer and the terms allowed logging. An English tribunal has said pasting a client letter into a public AI tool puts it in the public domain. Warn clients.

Can AI handle cross-border succession questions?

It is a reasonable place to start issue-spotting, but the accuracy data are poor. Benchmarks show general and legal AI dropping 11–14 points on multi-jurisdictional questions, and a 2025 study found frontier models hallucinating on more than half of location-specific legal questions. Use AI to frame the questions, then take the answers from local counsel or primary sources.

Put this into practice

AI Lab for Lawyers: a live, hands-on course for practising lawyers

4 live sessions (8 hours), recordings included, taught by Dr. Niklas Schmidt of Wolf Theiss. Rated 4.7/5 from 381 ratings; alumni from Norton Rose Fulbright, Paul Hastings, Pillsbury and Taylor Wessing.

Reserve a seat · €200 Maven money-back guarantee · certificate of completion

Practice Areas

AI for Tax Lawyers and Tax Advisers: Research, Rulings and Returns

AI for tax lawyers works where sources are supplied and numbers are checked. What tax authorities, Big Four and Blue J do, plus a worked withholding example.

10 min read

Guides

Which AI Tools Train on Your Data? A Confidentiality Checklist for Lawyers

Which AI tools train on your data, plan by plan as of September 2026: ChatGPT, Claude, Gemini, Copilot, legal vendors, plus bar rules and a 10-point checklist.

11 min read

Regulation & Ethics

Attorney–Client Privilege and AI: Heppner, Warner, the NYT Logs and Public Chatbots

Consumer chatbot memos are not privileged. US v. Heppner, Warner v. Gilbarco and the 20-million-log OpenAI order, plus rules that keep privilege intact.

10 min read

Playbooks

Extract Data from PDFs and Scans with AI: Contracts, Invoices and Court Files

How lawyers extract data from PDFs and scans with AI: OCR first, a fixed schema, a full lease-extraction prompt, batch runs in Cowork or n8n, and sampling.

11 min read

Use Cases

AI Meeting Notes and Transcription for Lawyers: Consent, Privilege and Setup

AI meeting notes are ethical for lawyers only with all-party consent, a no-training data path and a retention rule. Bar opinions, tools and a file-note prompt.

8 min read

Regulation & Ethics

Do You Have to Tell Clients You Use AI? Disclosure Rules and Practice

When lawyers must disclose AI use to clients under ABA 512, California, CCBE and ÖRAK rules, when consent is required, and a sample engagement-letter clause.

10 min read