In short
- Good candidates are steps that follow a pattern: sorting incoming mail, reading documents, preparing deadline entries, drafting routine letters and summarising files.
- Legal judgement, strategy, advice and every filing stay with the lawyer. The AI prepares drafts, the lawyer approves.
- Confidentiality comes first: a contract with the provider, processing in the EU or UK, no client data in consumer AI accounts and a written AI policy for the firm.
- Prospective clients increasingly ask AI assistants which firm to contact. Firms with clear practice-area pages, consistent profiles and register entries are easier to name.
Typical workflows in a law firm
These are typical processes, not client projects. The pattern is the same in each: something comes in, the AI prepares, a person checks and approves, and the result goes into the practice system with a log entry.
1. New enquiries and matter opening
Today an enquiry arrives by email, web form or telephone note. Someone reads it, decides which department it belongs to, checks for conflicts and opens the file.
- Input: enquiry email with attachments, web form entry or forwarded message.
- AI prepares: classifies practice area and urgency, extracts parties, dates and references, looks up existing clients and opposing parties in the practice system and drafts the reply asking for missing details.
- Lawyer checks: the conflict check itself, the decision to accept the instruction and the engagement letter.
- Output: a matter-opening form filled in as a draft and a reply that goes out only after approval.
2. Incoming post and deadlines
Letters from courts, authorities and the other side arrive on paper, by email and through court e-filing systems. Someone reads each one, files it and notes every deadline.
- Input: scanned or electronic documents.
- AI prepares: recognises the type of document, the court or authority, the reference number and any stated deadline, matches it to the matter and proposes the diary entry with the date and the passage it comes from.
- Lawyer checks: every deadline against the document. Calculating time limits remains a legal task.
- Output: the document filed, the diary entry and a short summary for the fee earner.
3. Document review against a checklist
Contracts, bundles and disclosure sets are read page by page, often for the same questions.
- Input: a contract, a bundle or an export from a data room.
- AI prepares: extracts parties, dates, notice periods, termination rights and unusual clauses against a checklist the firm defines, and lists every finding with the page and the quoted passage.
- Lawyer checks: each finding against the source text, then makes the assessment and gives the advice.
- Output: a review table or the draft of a summary memo, with page references.
4. First drafts of routine documents
Engagement letters, client updates, standard letters and attendance notes follow templates but are still typed from scratch.
- Input: the file, the firm’s templates and precedents, a dictated or recorded note.
- AI prepares: fills the firm’s own template from the file data and drafts the client update or attendance note.
- Lawyer checks: reads, edits and signs off. Nothing is sent automatically.
- Output: a draft in the firm’s own format, saved in the document management system.
What stays with the lawyer, and what we do not do
- Advice and judgement: assessing the merits, advising the client and negotiating. The AI gives no legal advice to clients.
- Filings, signatures and sending: nothing is filed, signed or sent without a lawyer’s approval.
- Citations: every case, statute and quotation an AI produces is checked against the original source. Courts in several countries have criticised filings with invented authorities.
- Decisions about people: we do not build systems that assess or select individuals, such as applicant screening. These count as high-risk under Annex III of the AI Act, see risk categories.
- Public chat without hand-over: a chatbot on the firm’s website answers organisational questions only, never legal ones, and says clearly that it is an AI.
Professional rules and data protection
Confidentiality is the core duty everywhere. The table lists the main sources by country as an orientation.
| Country | Basis | What applies in essence |
|---|---|---|
| England and Wales | SRA Code of Conduct for Solicitors, paragraphs 3.5 and 6.3, and the SRA report on AI in the legal market | Client affairs stay confidential unless disclosure is required or permitted by law or the client consents. The firm remains accountable for work done on its behalf, including work prepared with AI |
| Ireland | Law Society of Ireland, Guidelines on generative AI | Responsibility for the output stays with the solicitor, who verifies it independently and protects confidentiality and privilege. The guidelines apply the Solicitors’ Guide to Professional Conduct to generative AI |
| Germany | § 43a (2) BRAO, § 43e BRAO, § 203 StGB and the BRAK guidance on AI | Duty of confidentiality. Using service providers requires a written contract and a commitment to secrecy. The BRAK expects lawyers to check AI results themselves and to review confidentiality before using AI tools on a matter |
| Austria | § 9 RAO | Lawyers must keep confidential what has been entrusted to them or has become known to them in their professional capacity, as far as the client’s interests require |
On top of this, the GDPR applies in the EU and, in a similar form, in the UK. In practice that means a data processing agreement with the provider (Art. 28 GDPR), processing in the EU or UK, no client data in private or consumer AI accounts and a written AI policy for the firm. Art. 4 of the AI Act requires staff to have sufficient AI literacy, and Art. 50 requires AI chatbots to identify themselves as such. This overview is general information, not legal advice. Check the current texts with your regulator or bar.
Systems you probably use
Typical practice management systems are Clio, LEAP, Actionstep and Proclaim in the UK and Ireland, and RA-MICRO, ADVOWARE and DATEV in German-speaking countries. Documents often live in iManage, NetDocuments or SharePoint. We connect through the interfaces these systems offer (API, export, email or a watched folder) and work alongside them where none exists. The names are for orientation only and do not imply any partnership.
How AI assistants recommend law firms
A prospective client asks ChatGPT, Gemini or Google’s AI Overviews for “an employment solicitor in Leeds”. The answer is assembled from public sources: the firm’s own pages, the Google Business Profile, regulator and law society registers, legal directories and reviews. Firms that state their practice areas, location and people clearly, and say the same thing everywhere, are easier to name.
- One page per practice area: typical situations, the process and the people responsible.
- The same details everywhere: name, address and practice areas identical on the website, the Google Business Profile, the register entry and directories.
- Structured data: Schema.org
LegalServicewith name, address and areas. - Factual content: answers to real client questions without promises of success, in line with the advertising rules.
Our article Law firm SEO covers the details, including the advertising rules by country.
Is it worth automating? Eight questions
- Does someone in the firm sort or re-key incoming documents every day?
- Are deadlines read from documents and entered by hand?
- Do you review similar contracts or bundles again and again against the same checklist?
- Are routine letters and attendance notes written from scratch although the content is already in the file?
- Does the same information live in email, the practice system and spreadsheets?
- Do enquiries wait because nobody has time for the first reply?
- Is there a written rule on what may go into AI tools and who checks the result?
- Can you name one task where a draft that needs checking would already save time?
Three or more yes answers usually point to a workflow worth testing. Question 7 is the prerequisite: without it, start with the AI policy.
Sources
- England and Wales: SRA, Code of Conduct for Solicitors, RELs, RFLs and RSLs and Risk Outlook: the use of AI in the legal market
- Ireland: Law Society of Ireland, Guidelines for the Use of Generative AI by the Legal Profession in Ireland
- Germany: § 43a BRAO, § 43e BRAO, § 203 StGB, BRAK: Hinweise zum Einsatz von KI (in German)
- Austria: § 9 RAO
- Regulation (EU) 2016/679 (GDPR), Art. 28: EUR-Lex; Regulation (EU) 2024/1689 (AI Act), Art. 4, 50 and Annex III: EUR-Lex