AI client intake at a law firm means using AI to prepare the summaries, lists and drafts that intake staff and lawyers then check, from the first inquiry to the opening of the matter. Most writing about AI client intake at a law firm is about converting more leads for personal injury practices. A business law firm with 50 to 300 people has a different intake: a conflict check, corporate information, an engagement letter and a new matter. This article takes that process step by step and sets out what ABA Formal Opinion 512 and California’s 2026 guidance ask at each point.
This article is not legal advice. ABA opinions interpret the ABA Model Rules; each state adopts its own rules and guidance.
Intake at a business law firm
In a transactional, employment or real estate practice, intake is mostly reading and organizing: who the client is, who the other parties are, what the matter involves and which documents exist. Volume is lower than in a consumer practice and each file carries more information about third parties. That shapes where AI helps and where it creates risk.
The tools are usually already in place. Firms on Microsoft 365 or Google Workspace can do most of what follows without a new intake platform.
The AI client intake steps at a law firm, one by one
A business law firm’s intake has five steps where AI can prepare work: the first inquiry, collecting facts and documents, the conflict check, the engagement letter and opening the matter.
First contact and the inquiry email
A prospective client writes or calls. AI can summarize the inquiry into a short note for the responsible partner: parties named, type of matter, deadlines and documents mentioned. The decision to respond and how stays with a lawyer.
From this inquiry email, list: the prospective client’s full legal name, every other person or company named, the type of matter, any deadline mentioned, and documents attached or referred to. Use only what is in the email.
Collecting facts and documents
Once the firm decides to proceed, staff send a questionnaire and request documents. AI can turn the answers and attachments into a structured summary and a list of what is missing, using the firm’s intake checklist.
Conflict check
AI can prepare the list of names to search: the client, affiliates, officers, counterparties and their counsel as they appear in the documents. The search runs in the firm’s conflicts system, and the conflict decision stays with the lawyer responsible for it.
Engagement letter
Starting from an approved template, AI can draft the scope and parties for a new letter. If the firm uses AI on the matter, the letter may also need to say so; see AI disclosure in engagement letters.
Opening the matter
The structured summary from the earlier steps fills the new matter form. Staff check each field against the source documents before the matter is opened.
Keeping intake data in the firm’s systems
Every step above works on files that already sit in the firm’s email, document management or Microsoft 365 or Google Workspace environment. Keeping the AI work inside those same systems, under the firm’s enterprise terms, avoids copying a prospective client’s information into a separate tool. If a step seems to need a new tool, that is the point to stop and check its terms, its access settings and whether the use would require client consent, before anyone tries it on a live inquiry.
What ABA Formal Opinion 512 asks at each step
ABA Formal Opinion 512 asks for informed consent where client information enters a self-learning tool, a reading of the tool’s terms, firm policies and staff training, and verification of every output.
Confidentiality and informed consent (Rule 1.6)
The opinion says “a client’s informed consent is required prior to inputting information relating to the representation into such a GAI tool” where the tool is self-learning, and that “merely adding general, boiler-plate provisions to engagement letters purporting to authorize the lawyer to use GAI is not sufficient.” At intake this applies from the first email: information about a prospective client’s matter is sensitive before any engagement exists.
Enterprise suites describe their terms in public documents. Google states that “Workspace does not use customer data for training models without customer’s prior permission or instruction” (Workspace privacy hub). Microsoft says Copilot “only surfaces organizational data to which individual users have at least view permissions” (Microsoft Learn), which matters for ethical walls between matters: if a user cannot open a matter folder, Copilot will not bring it into that user’s answers.
Reading the terms of use of the tool
ABA 512 says “all lawyers should read and understand the Terms of Use, privacy policy, and related contractual terms and policies of any GAI tool they use.” California’s 2026 guidance adds that “Reasonable efforts require more than reliance on generalized marketing assurances.” For intake, that means someone at the firm reads the actual terms for the product and plan the firm uses, and records the answer to two questions: is client information used for training, and who can access it.
Policies and training for staff (Rules 5.1 and 5.3)
Intake is often run by staff. The opinion says “Managerial lawyers must establish clear policies regarding the law firm’s permissible use of GAI, and supervisory lawyers must make reasonable efforts to ensure that the firm’s lawyers and nonlawyers comply with their professional obligations when using GAI tools,” and that supervision includes making sure staff are trained. A one-page intake procedure that names the approved tool, the permitted steps and the reviewer is a practical form of that policy.
Checking the output
Texas Opinion 705 says lawyers “should always verify the accuracy of any responses received from a generative AI tool.” At intake the most important checks are the names that go into the conflict search and the facts that go into the engagement letter.
When AI acts on its own: California’s 2026 guidance on agentic tools
California’s 2026 guidance says a lawyer must not let an agentic AI system send client information outside the firm on its own without appropriate safeguards and human review. Some tools now act without a person in between, for example by reading the inbox and replying, or by filing documents. The California guidance treats this separately. It says agentic systems with access to email and files raise “acute confidentiality considerations under rule 1.6,” and that “A lawyer must not deploy an agentic AI system in a manner that permits autonomous external transmission of client information, including automated communications, filings, or data transfers, without appropriate safeguards and human review.”
For intake, our reading is simple: AI can prepare drafts and lists inside the firm, and anything that leaves the firm, such as a reply to a prospective client, is sent by a person.
A one-page intake procedure
The policy that ABA 512 expects does not need to be long. For intake, one page that staff can follow is usually enough, and it is easier to keep current than a general AI policy. It can cover:
- The approved tool for intake, the plan or license it runs under, and the date someone at the firm read its terms.
- Which intake steps may use it: summarizing the inquiry, structuring questionnaire answers, preparing the list of names for the conflict search, drafting the letter from the template.
- What may never be entered, for example information from a prospective client before the firm has decided to proceed, if counsel advises that.
- Who checks each output, and what they check: names and entities against the documents, facts against the questionnaire, the letter against the approved template.
- That every message leaving the firm is sent by a person.
A short training session can then walk staff through the page with one real, redacted example from your own intake, which is the kind of supervision and training Rules 5.1 and 5.3 describe.
From the attached intake questionnaire and the documents listed, prepare a list of every person and entity to run through our conflicts system: the client, parent and subsidiary companies, officers and directors named, counterparties and their counsel. Give the source document for each name. Do not judge whether a conflict exists.
What to keep with people
Five intake decisions stay with people, and they are also the points a prospective client expects a lawyer to handle personally, which makes them easy to explain in the engagement letter and in the intake procedure:
- the decision to take on a client or matter
- the conflict decision and any waiver
- advice of any kind, including in the first reply
- the final engagement letter and its fee terms
- every message that leaves the firm
Everything else in intake can be prepared by AI and checked by a person. Onboarding is one of the processes we suggest as a first candidate in Choosing the first process to review, and a review can run on redacted samples as described in A review without access to your systems.
Frequently asked questions
Can law firms use AI for client intake?
ABA Formal Opinion 512 does not prohibit it. It sets conditions: informed consent before inputting client information into self-learning tools, reading the tool’s terms, firm policies and staff training, and verifying output.
What does ABA Formal Opinion 512 say about confidentiality?
It says a client’s informed consent is required before inputting information relating to the representation into a self-learning generative AI tool, and that general boilerplate provisions in engagement letters are not sufficient.
Do lawyers need client consent to use generative AI?
For self-learning tools, ABA 512 requires informed consent before inputting information relating to the representation. California’s 2026 guidance says a lawyer must not input confidential information into a tool that may present material risks to confidentiality absent informed consent. Check your own state’s rules.
Can AI run a conflict check?
AI can prepare the list of names to search from the intake documents. The search runs in the firm’s conflicts system and the conflict decision stays with the responsible lawyer.
Are ABA formal opinions binding?
ABA formal opinions interpret the ABA Model Rules. Each state adopts its own rules of professional conduct and issues its own guidance, so check your jurisdiction.
Do paralegals and staff need AI training under the ABA rules?
ABA 512 says supervisory lawyers must make reasonable efforts to ensure that lawyers and nonlawyers comply with their obligations when using generative AI, and that supervision includes making sure staff are trained.
Sources
- ABA Standing Committee on Ethics and Professional Responsibility, Formal Opinion 512, Generative Artificial Intelligence Tools, July 29, 2024 (PDF)
- Google Workspace Admin Help, Generative AI in Google Workspace Privacy Hub
- Microsoft Learn, Data, Privacy, and Security for Microsoft Copilot
- State Bar of California, Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law, 2026 version (PDF)
- Professional Ethics Committee for the State Bar of Texas, Opinion 705, February 2025
This article is general information. It is not legal or tax advice. For your own situation, ask your counsel, your tax adviser or the professional body that governs your practice.