Hamid Kohan, President & CEO, Legal Soft, author of three books on law firm growth, & a frequent keynote speaker on AI and legal innovation.
getty
Picture a potential client who calls a law firm at 8:40 p.m. after a car accident. Nobody picks up, so she dials the next firm on her list.
That scenario is common. In a study Clio commissioned, researchers contacted 500 law firms by phone and email. Only 40% picked up when called, and 48% were unreachable by phone even after the firm had a chance to respond to messages.
It is no surprise that firms are looking to AI to answer the phone. I have called intake one of the safest places to start with AI, and I still believe that. But safe is not the same as duty-free. Here is what I think firm leaders should weigh before the line goes live.
Professional Duties Start Before The Contract
The American Bar Association’s “Formal Opinion 512” addresses lawyers’ use of GenAI tools in general, and though not specific to phone lines (and firms should still confirm how their own state bar applies these rules), I find much of the guidance and logic carries over to intake.
The opinion tells lawyers to weigh the risk that information will be disclosed to or accessed by people outside the firm. It calls for a client’s informed consent before information from a representation is entered into a self-learning tool, and it says general boilerplate in an engagement letter is not enough.
At intake, there is no engagement letter yet to carry any consent. The opinion also notes that Rule 1.18(b) extends similar confidentiality protection to prospective clients’ information, and a caller describing an injury, an arrest or a divorce is handing over exactly that.
In my view, that makes three questions practical rather than theoretical: where call audio and transcripts are stored, who can open them and whether a vendor can use them to train its models.
What Callers Expect When AI Answers
Gartner surveyed 3,566 customers in early 2026 and found that 87% say it is essential for companies using generative AI to offer a way to reach a human. Gartner analyst Eric Keller advises service leaders against making AI a mandatory first step for every issue.
Pew Research Center found that 72% of U.S. adults say it is extremely or very important that a healthcare provider tells them if AI is used in their care, and 56% say they should be told when AI is used to schedule an appointment. Of course, the survey concerned healthcare, not law. Even so, a person calling about an arrest or a custody dispute is just as likely to be anxious, and a voice that sounds human but is not can feel like deception, whatever the law requires.
The Rules Are Arriving Unevenly
Disclosure rules are emerging state by state. Utah requires individuals in regulated occupations to disclose generative AI use verbally at the start of high-risk interactions, a category that includes legal advice. Maine prohibits using chatbots, including voice ones, to mislead consumers into thinking they are dealing with a human.
Neither directly includes law firms, but I would not wait for a regulator or a court to decide. My recommendation is simple: Say at the start of the call that the caller has reached an AI assistant, and make the path to a person obvious.
Replacing People Is Not The Safe Shortcut
It is tempting to fund the technology by trimming the front desk. The research suggests caution. In findings published in December 2025, Gartner reported that over 80% of organizations expected to reduce customer-service agent headcount within 18 months, mainly through attrition. Yet only 20% of customer-service leaders surveyed that October said they had actually reduced agent staffing because of AI, which Gartner reiterated in February 2026.
That said, I am not arguing against AI. I believe the stronger design is AI that gathers information and handles routine questions, with people ready to take over when a caller is distressed, confused or asking for advice.
Who Supervises It?
Opinion 512 says managerial lawyers must establish clear policies on the firm’s permissible use of generative AI, and that supervisory lawyers must make reasonable efforts to ensure that lawyers and nonlawyers comply with their obligations when using it. An AI phone line should not be the one tool that runs without a policy. I would put four questions in writing:
1. What may the AI say, and what must it never say?
2. When must it hand the call to a person?
3. Who reviews transcripts?
4. How often is that review done?
Before The Line Goes Live
None of this argues for leaving the phone unanswered, because a 40% pick-up rate carries its own risk. It argues for deciding deliberately. Before an AI line goes live, a managing partner should be able to answer three questions: who hears the call, does the caller know and who is watching. If any answer is “I’m not sure,” the front door is not ready.
The call can be answered by software. The duty to the caller cannot.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?
