The case that called after hours went to whoever picked up.

Yes. A single recovered intake call that converts to a retained case covers 6 to 12 months of AI receptionist cost. The break-even is one booked consultation per month, and most law firms miss more than one intake call per week to voicemail or after-hours silence.

Here is the math for a personal injury firm. Average case value: $5,000 to $15,000 in attorney fees (contingency). AI receptionist cost: $249/mo ($2,988/yr). If the AI books just one PI case per quarter that would have gone to voicemail, the annual ROI is 5x to 20x.

For a family law practice charging $350/hr with a $5,000 retainer, one retained client per month is $60,000 a year in new revenue from a $2,988 annual cost. The ROI math works for every practice area where the average case value exceeds one month of AI receptionist cost.

The hidden cost of not answering is the caller who never calls back. According to industry research, most first-time callers who reach voicemail hang up and call the next firm on the list. They are not invested in your firm yet. They are shopping, and silence tells them you are closed.

An AI receptionist is a front-desk tool, not a paralegal, not a junior associate, and not a substitute for attorney judgment. Here are the real limitations, because a vendor that hides failure modes is not one you should trust with your intake line.

• It cannot give legal advice. It will not and should not assess the merits of a case, recommend a course of action, or make representations about outcomes. The attorney-client relationship begins with the attorney, not the phone system.
• It cannot handle highly emotional callers the way a trained human can. A caller in crisis (domestic violence, custody emergency, criminal arrest) may need human empathy and nuance that an AI voice cannot fully replicate. The AI can recognize urgency cues and warm-transfer immediately, but it cannot replace a compassionate human in a high-stakes emotional moment.
• It cannot navigate attorney-client privilege boundaries. The AI collects intake information. Whether that information is privileged, how it is stored, and who has access is the firm's responsibility to configure. AI receptionist providers should offer SOC 2-compliant storage and clear data-handling agreements, but the privilege analysis is the attorney's call.
• It struggles with heavy accents and unusual speech patterns. Current AI voice technology handles standard American English and Latin American Spanish well. Heavy regional accents, speech impediments, and code-switching mid-sentence can cause comprehension drops. The AI should be configured to transfer to a human when it detects low confidence.
• It cannot handle consultative intake where the call IS the service. If your practice model involves a paid initial consultation where the first call is substantive legal work, an AI receptionist books the appointment but does not replace the consultation itself.
• Low call volume may not justify the cost. A solo practitioner who gets 5 calls a week and answers them all personally does not need an AI receptionist. The ROI requires enough missed or after-hours calls that at least one per month would have converted to a paying client.