AI intake is software that answers, qualifies, scores, and follows up with every lead the moment it arrives, 24 hours a day. It fixes the two things a human intake team physically can’t deliver: instant response and coverage at 9pm on a Friday.
I built my firms around intake. Not better marketing, not better litigation. Intake. My flagship firm handles lemon law cases in all 50 states, and after nine years of running consumer firms, the highest-return thing we ever fixed was how fast we answer.
Most firms never fix it. They spend $50,000 a month on ads and then let the leads those ads generate sit in a voicemail box until Monday.
The math nobody wants to run
The Lead Response Management study out of MIT found that calling a lead within 5 minutes instead of 30 makes you about 21 times more likely to qualify them. Not 21 percent better. 21 times.
Our own numbers say the same thing. A lead we reach in the first minute or two signs at several times the rate of a lead we reach the next morning. By morning, that person has filled out three other firms’ forms. Whoever calls first usually wins, and it’s almost never about who has the better firm.
Then there’s coverage. Legal leads don’t keep business hours. The person whose transmission died on the freeway is searching at 9pm. The employee who got fired is searching Saturday morning. If your intake runs 9 to 5 on weekdays, you’re staffed for roughly a quarter of the hours your leads actually exist.
Before we rebuilt our system, I wrote down everything wrong with our old intake. Top of the list: qualified leads waited hours, sometimes days, before an intaker called them. We leaned too hard on text and sat waiting for replies when a three-minute call would have qualified the lead on the spot. And nobody got notified the moment a lead qualified, so hot leads cooled in a queue. None of those were staffing problems. They were system problems, and every one of them cost signed cases.
What AI intake actually is
Strip away the hype and AI intake is five functions:
- Answering every inquiry 24/7, whether it arrives by phone, web chat, or text
- Asking your qualification questions and checking answers against your case criteria
- Booking qualified leads straight onto your intake team’s calendar
- Logging everything to your CRM automatically, with no retyping
- Handing the lead to a human the moment it’s worth a human’s time
That last one matters most. AI intake done right is a filter and an accelerant. Humans still sign the clients.
The three layers of the stack
Layer one is AI voice. An AI receptionist answers the calls that would otherwise hit voicemail, holds a real conversation, asks your screening questions, and books or transfers. Smith.ai runs an AI receptionist alongside its human answering service. Aloware, which my firms use for calling, has AI agents built into the dialer. Purpose-built legal tools like OpenIntake exist for exactly this job. I break the whole category down in the voice AI guide.
Layer two is chat and SMS qualification. A website chatbot plus a text bot that engages the second a form comes in. At my firms, the first text goes out within seconds of a form fill: a short message from a named assistant asking about the vehicle and whether the lead has a few minutes for questions. Reply rates on that instant first text embarrass every email sequence we’ve ever run.
Layer three is the one almost nobody builds: AI lead scoring. Every lead that comes into my lemon law firm gets scored against our case criteria before a human touches it. For lemon law, that means three questions. Is the vehicle under warranty? How many repair attempts? How many days out of service? The AI reads the intake answers and sorts the lead into a status. Qualified means an intaker calls in under 5 minutes. Escalation or uncertain means a human reviews first. Junk never reaches a human at all.
How it runs at my firm
CPLG targets 800 sign-ups a month. The intake team, run by a full-time intake manager, does not dial cold lists or babysit chat windows. AI qualifies and scores first. Intakers spend their day on leads the system already marked qualified, which means every call they make has a real chance of becoming a client. The retainer conversation stays human.
Follow-up is where most firms quietly die, so we automated all of it. A lead who doesn’t answer doesn’t get a sticky note on someone’s monitor. They enter a sequence: calls and texts on a schedule until they respond or clearly opt out. This is agent territory, and the AI agents guide covers how sequences like this get built. Same after the retainer goes out. Automated reminders run until the document is signed, and my settlement pipeline starts with clients other firms would have lost at hello.
Notice the flip. Most firms have humans doing coverage badly and nobody doing scoring at all. We have AI doing coverage perfectly and humans doing only the thing humans are actually good at.
What stays human
Two things I refuse to automate. First, the empathy call. Someone whose car has been in the shop for four months is angry and tired of being dismissed. Their first real conversation with the firm needs a person who can listen and mean it. Second, the retainer close on big cases. When the case is worth serious money, a human closes it, every time.
AI buys you speed and coverage. Humans build trust. Confuse those two jobs and you’ll lose cases from both directions.
Where to start if you’re small
Don’t try to build my stack on day one. The order that works:
First, missed-call text-back. Any missed call triggers an instant text: “Sorry we missed you, can you tell me a bit about your case?” It’s the cheapest fix in legal marketing and it runs by tonight.
Second, after-hours AI answering. Keep humans on the daytime phones if you like them there, but let an AI receptionist take nights and weekends. That alone roughly quadruples your coverage hours.
Third, scoring. Write down your case criteria, the way I did with warranty status, repair attempts, and days out of service. Then have AI check every lead against them before your staff sees it. If you already draft with AI daily, this is a prompt away. If not, start with the prompt library and Claude first.
What it costs
| Layer | What it does | Typical cost |
|---|---|---|
| Missed-call text-back | Instant text after any missed call | Under $100/mo, often free in your CRM |
| AI voice receptionist | Answers, screens, and books 24/7 | $100 to $300/mo at solo call volume |
| Chat and SMS bots | Engage web leads within seconds | Same range |
| AI lead scoring | Sorts every lead against your case criteria | Custom build, a one-time project |
A fully custom scoring system like ours is a real project, though far cheaper to build now than it was even a year ago.
Price it against the alternative. One signed case that would have gone to voicemail covers a year of any of these tools. In lemon law, one case covers several years.
The bottom line
- Speed-to-lead beats every other intake metric
- AI takes answering, qualifying, scoring, and follow-up
- Humans keep the empathy call and the retainer close
- Missed-call text-back is the first fix, and it runs by tonight
Your homework this week
Call your own firm at 7pm on a Friday. Better yet, fill out your own contact form on Saturday morning and start a timer. If the answer is voicemail and the timer is still running Monday, you found your leak, and it’s costing you more than any tool on this page. Fix the leak first. The roadmap covers what comes after.