For a small law firm, the most useful place to start with AI is the front office: answering calls after hours, collecting documents, drafting routine client updates, summarizing consults and getting hours onto the bill, with an attorney approving anything that goes out. What it should never do is give legal advice, decide a conflict, or handle client information in a tool nobody has checked for confidentiality.
Most of what you’ll read about AI for law firms is about legal research and contract review tools. This post is about a different part of the firm: the phone, the chase for documents and the same status email written for the tenth time. That’s the work our AI integration service is built around, and it’s what this post covers.
Where AI earns its keep in a small firm’s front office
Each of these is a job AI can do well today, as long as a person at the firm owns the result.
| Front-office job | What AI can do | What stays with a person |
|---|---|---|
| After-hours and overflow calls | Answers in your firm’s name, takes down the basics of the matter, and books a consult | Legal advice, whether you take the case, the conflict check |
| Document collection | Keeps a checklist per client, sends reminders and tells you when the file is complete | Deciding what the matter actually needs |
| Routine client emails and status updates | Writes a first draft from the file, in your firm’s voice | Reading, editing and sending it |
| Consult and meeting summaries | Writes up decisions and next steps, with everyone’s permission to record | Checking the summary before it goes in the file |
| Long document summaries | Pulls out the key points, each tied to the page it came from | Verifying every point before anyone relies on it |
| Time to billing | Puts hours logged against a matter onto the invoice draft | Reviewing the invoice before it goes out |
The phone is often where a small firm feels it first. A prospective client who reaches voicemail at 7 p.m. may simply call the next firm on the list. An AI receptionist can pick up, say plainly that it’s an automated assistant, take down what your firm has decided it needs, and put a consult on the calendar. It doesn’t give an opinion on the case, quote fees it hasn’t been given, or tell anyone the firm will represent them. Opinion 512 (more on it below) notes that the rules extend similar protections to prospective clients’ information, so the call deserves the same care as the file. If you’re weighing a human answering service against AI, we compared the options in answering service vs AI receptionist.
The lines AI shouldn’t cross at a law firm
Whatever a firm builds or buys, these are the decisions we’d hold the line on:
- No legal advice, to anyone. Not to a caller, not in an email draft that goes out unread.
- No conflict decisions. AI can collect the names your staff need. A person at the firm runs the check and makes the call.
- Nothing reaches a client or a court without an attorney’s review. AI drafts. A lawyer decides.
- Every person and tool sees only what its role needs. The receptionist doesn’t need the whole matter file.
- Client information goes only into tools the firm has vetted. More on that below.
What the ethics opinions say
Two opinions are worth reading in full, and we’d rather point you to them than paraphrase loosely. We checked both on October 7, 2026.
The American Bar Association’s Formal Opinion 512, Generative Artificial Intelligence Tools, dated July 29, 2024, says lawyers using these tools “must fully consider their applicable ethical obligations.” In short:
- Competence. Lawyers need a reasonable understanding of what a tool can and can’t do, and must review its output to a degree that fits the task.
- Confidentiality. For “self-learning” tools that could reveal client information, the opinion says a client’s informed consent is required before information relating to the representation goes in. It adds that general, boilerplate language in an engagement letter is not enough, and that lawyers should read a tool’s terms of use and privacy policy.
- Communication. Depending on the facts, a lawyer may need to tell clients about AI use, including when a client asks, when consent is needed, or when it bears on the fee.
- Supervision. Managing lawyers must set clear firm policies on permissible AI use and train lawyers and staff. The same duty to vet outside providers applies to AI vendors.
- Fees. A lawyer billing hourly must bill actual time. That can include the time spent giving the tool the information and reviewing its draft. The opinion also says a lawyer “may not charge a client to learn about how to use a GAI tool or service that the lawyer will regularly use for clients.”
In Texas, the Professional Ethics Committee for the State Bar of Texas issued Opinion 705 in February 2025. It says a lawyer should be “reasonably satisfied that the program will not reveal confidential information to others or permit the use of such information to the disadvantage of the client,” and if not, should “at a minimum” not input confidential information “without client consultation and consent.” It also says a lawyer should “always verify the accuracy of any responses received from a generative AI tool,” and that a lawyer billing hourly “is never justified in charging a client for hours not actually expended.”
Other states have their own guidance, and these opinions can be updated. Read the current version for where you practice.
The confidentiality controls to insist on
Those duties sit with the firm. What a technology partner can do is build the system so the firm can meet them. On every system we build for a firm, we put these in place:
- Accounts the firm owns. The AI tools, the document portal and the data live in the firm’s accounts from day one, not ours and not an employee’s personal login.
- Role-based access. Staff see the matters and documents their role needs, and nothing else.
- An audit trail. You can see who opened what, and when.
- Your data kept out of public AI training. We set the tools up so client information isn’t used to train public AI models.
- Encryption in transit and at rest, plus export or deletion on request.
Opinion 512 also treats an AI provider like any outside vendor the firm has to vet. That includes us. Ask us how access works, where the data lives and what happens to it if you leave, and expect straight answers. We wrote up how we handle client information on our data handling page. If your staff are already pasting things into ChatGPT, read is ChatGPT safe for business data first.
Where we stop: we build systems, we’re not your lawyer
We build the technology. We are not your ethics counsel, and we’re not your lawyer. Whether a particular use needs client consent, what your engagement letter should say, and how your state’s rules apply are questions for you and, where it helps, your own ethics counsel. When something crosses into that territory, we’ll say so and stop there.
What we can do is make the careful path the easy one: attorney review built into the process, client information kept in accounts the firm controls, and a record of who touched what.
How to choose the first job for AI
Start small and pick one job. The best first candidate is usually the one that eats attorney time without needing attorney judgment, like after-hours calls or the document chase. Before anything is built, decide three things:
- What the tool may see, and what it may never see.
- Who reviews its work before it reaches a client.
- How you’ll explain it to clients if they ask.
Check what you already pay for, too. Many practice management platforms include intake forms, client portals or AI drafting. If yours already does the job well, use it. We’ll tell you when that’s the right answer. Our law and accounting page shows how the pieces fit together for a firm.
If you’d like a second opinion on where AI fits in your firm’s front office, and where it shouldn’t, that’s what the call below is for.