If you run a family law, criminal defense or probate firm, a lot of your callers are having one of the worst weeks of their lives. So when someone suggests letting an AI answer the phone, the worry is fair. Will it sound cold? Will it miss something important? Will it say the wrong thing to someone who's barely holding it together?
We build AI intake, so we have a stake in this. We'll give you the honest answer anyway. AI handles a lot of distressed callers well. It should not handle all of them, and the ones it shouldn't handle need a clear path to a person.
What "distressed" covers
Not every upset caller is in crisis. It helps to separate them.
- Upset but safe. A parent worried about custody, someone whose spouse just moved out, a family member calling after an arrest, an adult child whose parent just died. Emotional, sometimes tearful, but not in danger.
- Urgent. A hearing tomorrow, a deadline this week, someone who's just been served. Stressful, time-sensitive, legal.
- Unsafe. A caller who mentions abuse, threats, fear for their kids, or thoughts of self-harm.
Most distressed calls to a small firm fall into the first two groups. That matters for how you think about AI.
Where AI does well
It stays calm. An AI agent doesn't get flustered when someone's yelling, and it doesn't get short with the fifth crying caller of the evening. Its tone at 11 p.m. is the same as at 11 a.m.
It's consistent. It asks the same intake questions in the same order every time. It doesn't skip the "is it okay to text you?" question because the caller seemed rushed. Human intake teams are good, but they get tired, distracted and busy.
It never sends a caller to voicemail. For someone who finally worked up the nerve to call a lawyer at midnight, a real conversation beats a beep. That alone changes the experience.
It doesn't give advice. A well-built intake agent is set up to steer legal questions back to the attorney every time. Well-meaning human staff sometimes slip and guess, especially when a caller is begging for an answer.
It gives a clear next step. Distressed callers want to know what happens now. Booking a consult on the call gives them something solid to hold onto.
Where AI falls short
It doesn't read people like a skilled human does. An experienced intake coordinator can hear the hesitation that means "I can't talk freely" or the flatness that means something is very wrong. AI is getting better at picking up cues, but it shouldn't be the last line of defense on subtle signals.
It shouldn't counsel anyone. Comforting words are fine. Crisis support is a job for trained people.
It can't make judgment calls the firm hasn't defined. If your firm hasn't decided what counts as urgent, the AI can't decide for you. Neither can a new hire, to be fair.
Some callers just want a person. That's a legitimate preference, and the agent should respect it.
The rule: safety calls go to a person
This is the part we won't bend on. Any call that touches safety needs a person. That includes:
- A caller who says they or their children are in danger.
- Any mention of domestic violence, threats or stalking.
- Any mention of self-harm.
- A caller who asks for a human.
For these calls, the agent should do a few specific things and then hand off.
- If the caller is in immediate danger, tell them to call 911.
- If they mention self-harm, give them the 988 Suicide & Crisis Lifeline (call or text 988).
- Ask whether it's safe to talk, and get a safe number and time to call back.
- Skip automated texts unless the caller says texts are safe.
- Route the call to a person at the firm right away, following the firm's written protocol.
We go deeper on steps 3 and 4 in safe intake for domestic violence callers.
What a good handoff looks like
A handoff that dumps a caller into a voicemail box isn't a handoff. Decide before you go live:
- Who gets escalated calls during the day, after hours and on weekends.
- How they're reached and how fast they respond.
- What information travels with the lead, especially the safe contact plan.
- What the agent tells the caller about what happens next.
"I'm going to connect you with someone from the firm now. If we get cut off, is this a safe number to reach you?"
We wrote a fuller guide on drawing this line in when callers need a person: where AI intake stops.
Questions to ask any AI intake vendor
If you're evaluating AI for intake, ask these directly:
- How does the agent respond when a caller mentions abuse or danger?
- How does it hand off to a person, and how fast?
- Can automated texts be turned off for a specific lead?
- Can we write and change our own escalation rules?
- What does the agent say when a caller asks for legal advice?
Vague answers to these are a warning sign.
So, can it?
Yes, for most distressed callers. AI is calm, consistent and always there, and for an upset person calling at midnight that's a big improvement over voicemail. No, for safety calls. Those belong with a person, and the system should be built to get them there fast.
The best setup for most small firms is a mix. AI answers every call and handles routine and upset-but-safe intake. People handle escalations and the calls that need judgment. That's the model in hybrid AI and human intake for law firms.
Where IntakeGenius fits
Our AI voice agent answers every call, runs your intake questions in a steady voice, and books the consult. When a call touches safety, the agent routes the call to a person at your firm, following rules we set up with you. If you'd rather have a dedicated person in the loop full-time, we also offer a trained intake specialist.
Want to hear how it handles a hard call? Ask us for a demo. It takes 15 minutes, no slides.