Lawyers & Law Firms · SoCal & nationwide
AI for Lawyers Who Can't Afford to Miss a Call
I spend a free day inside your law firm, watch where the calls and the hours actually go, and hand you a written plan of three to five AI moves with honest monthly costs. No software pitch, no build, no referral fees.
It’s quarter to six on a Tuesday in a three-lawyer office in Long Beach. The senior partner has been in mediation since ten that morning. Her phone shows six missed calls from numbers she doesn’t recognize, and the office line rolled to voicemail at five, same as every night. One caller was a woman who had just been served with an eviction notice that afternoon. She left no message. By the time the partner checks the voicemail box the next morning, that woman has already called someone else. “As a solo attorney, I can’t afford to miss calls,” she told me the week I spent in her office. “The phone rings during depositions, court, and client meetings, and after five it just hits voicemail and dies there.”
That’s what owners actually ask me about. Not the technology. They ask how to stop losing clients they already paid to attract, how to get an evening back once in a while, how to bill for the hours they actually worked, and whether any of this is safe in a profession where one made-up citation can end up in a sanctions order.
I’m Isaac Krupp. I spend one free working day inside a small business and hand the owner a written plan of three to five AI moves with honest monthly costs. I don’t sell software, I don’t build anything custom, and I don’t take a referral fee. What follows is what I would tell a friend who runs a law firm.
What AI for lawyers can actually do right now
Here are the problems that come up most often when I sit inside a small firm, and the plain off-the-shelf fix for each one.
Calls go unanswered, especially after hours
You are in court or a deposition, and the phone rings anyway, then rolls to voicemail after five until Monday. A 2025 audit of small and mid-sized firms found 34.8% of business-hours calls went completely unanswered, and 80% of the callers who reached voicemail hung up without leaving a message.
The fix is an AI receptionist in front of your existing number. It answers in the firm’s name, asks the intake questions you wrote, screens out sales calls, books a consultation on your real calendar, and texts you a summary, day or night.
It should never give legal advice, quote a fee, or decide who becomes a client. It takes information and books time, nothing else, and a distressed caller in a criminal or immigration matter still gets routed to a person.
New leads never get followed up
You paid for the ad, the contact form filled up, and the lead sat unread for two days. Research summarized by the Oklahoma Bar found 64% of prospective clients got no follow-up at all by phone or email, and the average firm spends roughly $649 per lead, and a separate study found only 33% of firms responded to a prospect’s email at all.
The fix is an intake pipeline: every form, chat, and voicemail lands in one queue, and an assistant sends an immediate, human-sounding acknowledgment in your firm’s voice, asks a few qualifying questions, and checks back if nobody has replied.
It should never negotiate a fee or discuss the substance of the matter. Its only job is keeping the conversation alive until a person takes over, and it stops the moment someone asks it to.
Admin work eats the billable day
You went to law school to practice law, and instead most of the day disappears into scheduling, forms, and email, so the real billable work happens after everyone else has gone home. Only about 3 of every 8 hours in a typical day are billable, with roughly 81% of the rest described as automatable admin, and that administrative load ranks as small firms’ top challenge year after year.
The fix is a short list of small automations: dictation that turns a voice note into a first-draft letter, and document assembly that fills your own forms from intake answers.
Nothing goes to a client or opposing counsel unread. Client files stay out of consumer chat tools whose terms allow training on what you type, and an attorney reviews the substance of every draft first.
Billable hours slip through the cracks
You know you did work you never wrote down, and by the time you reconstruct the day from memory, half the six-minute increments are gone. One survey found 63% of lawyers use time-tracking tools, but only 38% believe they bill for all the hours they actually work.
The fix is passive time capture built into your practice-management or billing software. It watches the calendar, email, and call log, proposes entries with a plain draft description, and you confirm or edit them at the end of the day instead of guessing a week later.
You still certify every entry yourself. The tool proposes, a person approves, because a padded or invented time entry is a fee dispute waiting to happen.
Rising costs and hard-to-fill staff roles
Your best paralegal leaves for a bigger firm paying more, it takes months to find anyone, and meanwhile you do paralegal work at a lawyer’s hourly rate. Small firms typically pay paralegals less than large firms do, and replacing one can cost 50% to 400% of salary, while a lateral hire commonly takes three to six months to become productive.
The fix is not replacing the paralegal, it is letting AI absorb the repetitive lookup and formatting work: summarizing discovery, assembling standard clauses from your own templates, and drafting routine correspondence for a person to check.
Supervision does not move. You answer for AI-assisted work exactly as you would for a paralegal’s, and sealed or medical-privacy material needs a tool vetted specifically for it, or it stays out.
Online reviews decide who gets hired
Clients rarely leave because of the legal result. They leave because you took two days to return a call, and then say so on Google before the next prospect even calls. A survey found 61.5% named responsiveness the single most helpful factor in choosing a lawyer, and 46.2% named a slow response a top deterrent, while secret-shopper research found 73% would not recommend the firm they had just contacted.
The fix is a simple review and response tool that asks a satisfied client for a review once a matter closes, and flags any client thread gone quiet for more than a day so someone can catch it early.
It should never write or post a response touching a client’s actual matter without your sign-off, and review requests never go to clients in criminal, family, or immigration cases.
Fear of AI inventing fake citations
You would love to draft faster, but you have read about lawyers fined for citing cases that do not exist, and you have no associate free to double-check everything by hand. The fear is earned: a tracked database has documented 486 cases worldwide, 324 in US courts, of AI-fabricated citations, and one 2025 case drew a $10,000 sanction over a ChatGPT-assisted brief with fabricated citations.
The fix is keeping two lanes separate: legal-specific research tools tied to a real case database return citations you can open and check, unlike a chatbot guessing at a case name.
Every citation still gets opened and read before it goes into a filing, and every brief runs through a citation checker before you sign it. You own every word that goes out under your name.
No time to vet or set up new tools
Every vendor promises their software saves ten hours a week, but you do not have the ten hours it would take to learn it and move your files over, so nothing changes. A 2026 survey found 27% of solos and 33% of small firms name finding the time as their biggest barrier to new technology, and more than half have no written AI policy at all.
The fix is a short guided setup, not another subscription to research alone: pick one narrow tool, connect it to the calendar or intake form you already use, and write a one-page policy for how staff are allowed to use it.
It should never involve moving privileged client files into a tool before its data handling and confidentiality terms have actually been checked.
A day in a law firm with AI turned on
Call her Priya, a solo family-law attorney in Riverside with one paralegal. Before AI, Priya is in court until nearly noon, and the office line rolls to voicemail because the paralegal is at the copier. Back at her desk she finds five voicemails, two of them possible new clients. She calls both back over lunch. One has already hired someone else. She starts an engagement letter for the other by opening the last one she sent and changing the names by hand. Three “any update?” emails wait until after her two o’clock hearing. At six she reconstructs the day’s time from memory and lands at, say, four billable hours out of nine.
After AI, every call that morning gets answered. The receptionist screens two new custody inquiries, takes exactly the details Priya told it to take, and books both for Thursday, texting her a summary of each. Back at her desk at noon, a drafted engagement letter is waiting, built from her own template and the intake answers; she changes one sentence about the fee and sends it in under ten minutes. The three status emails have replies drafted from the matter file, ready for her to approve. At five forty she confirms eleven proposed time entries instead of guessing. At, say, $300 an hour, the extra hour and a half she now captures most weeks is real money. Nothing about Priya’s judgment changed. The conflict check and the hard phone call to an anxious client are still hers. What moved was the retyping and the guessing.
What it costs, honestly
I never quote a vendor’s price, since it changes and I’m not selling any of it. These are the ranges I see inside small firms.
An AI receptionist usually runs around a couple hundred dollars a month. Intake tools, drafting assistants, and time-capture tools are often just tens of dollars per seat on top of software you already pay for, and legal research tools with verified citations sit at the expensive end, typically a few hundred dollars per attorney a month. A first plan usually lands at a few hundred dollars a month across the whole firm.
The expensive mistake is paying a developer to build custom software before anyone has used the off-the-shelf version for six months. Buy the boring tool first, and build only once you’ve proven it can’t do the job.
What NOT to hand to AI in a law firm
This is a regulated profession, and these lines aren’t a matter of taste. Your duties of confidentiality and competence, the supervision you owe non-lawyer staff, and your state bar’s own rules on advertising and generative AI govern every tool here. Nothing on this page is legal advice about how those rules apply to your practice.
Legal judgment stays with you. No tool decides whether to take a case, states its merits, quotes a fee, predicts an outcome, or opens a matter, and the conflict check is yours to run. Confidential facts go only into tools backed by a written agreement that keeps your data out of training and stored in the United States. Nothing gets filed or sent unread, every time entry is certified by a person, and trust accounting stays untouchable.
How the free AI day works for a lawyer
I come to your office for a full working day, in person anywhere in Los Angeles or Southern California, or by video anywhere in the United States. No pitch deck.
In the morning I watch the phones, the front desk, and the back office. I call your main line myself a few times, including once over lunch, and show you what a new client hears and how long it takes anyone to call back. By lunch I hand back the tasks that ate the most non-billable minutes.
In the afternoon we walk through your top pains together: last month’s calendar next to last month’s invoices, your public reviews next to your intake log, and one recent brief next to what a legal research tool would have drafted and needed you to verify. At the end of the day you get a written plan, three to five AI moves in order, with honest monthly cost ranges and a clear line around what each one must not touch. The how it works page has the full shape of the day.
If you also handle your own books or lean on a CPA, the AI for accountants page covers that side of a small practice, and AI for financial advisors shows what the other regulated office down the hall expects too.
You don’t need to become an expert in AI for lawyers. You need one day, one plan, and your evenings back. The point isn’t the software, it’s the hours you stop spending on intake and billing going back into the cases that actually grow the firm, with enough left to make it home for dinner. Book the free AI day.
Straight answers
Questions lawyers & law firms ask me
How much does ai for lawyers cost a small firm?
It depends on what you turn on first. An AI receptionist usually runs around a couple hundred dollars a month. Drafting and time-capture assistants are often just tens of dollars per seat, sometimes already inside software you pay for. Legal research tools with verified citations cost more, typically a few hundred dollars per attorney a month. Most firms start with a few hundred dollars a month total.
Is it safe to put client information into ChatGPT or other AI tools?
Not under a free consumer plan by default. Many free tiers let the vendor train on what you type, which is a confidentiality problem before it's a technology problem. Business-tier tools and legal-specific platforms usually offer agreements that keep your data out of training and stored in the United States. Read the terms, check your state bar's guidance on generative AI, and keep privileged strategy notes out of anything you haven't vetted.
Can ai for lawyers do legal research without inventing fake cases?
General chatbots have made up case names, and courts have sanctioned attorneys who filed them unread. Legal-specific research tools tied to a real case database return citations you can open and check, which is different from a chatbot guessing. Even then, write the rule down: every citation gets opened and read before it goes into a filing, and every brief runs through a citation checker before it's signed. You still own every word in it.
Will an AI receptionist give legal advice to callers?
It shouldn't, and a properly configured one won't. It answers in your firm's name, tells callers it's an automated assistant, asks the intake questions you wrote, books a consultation, and sends you a summary. It never quotes a fee, predicts how a case will go, says whether you'll take it, or runs a conflict check on its own. A caller in a criminal or immigration matter in distress should trigger a human callback, not a script.
Do I need a tech person to set up ai for lawyers?
For the moves that pay off first, no. Turning on an AI receptionist, connecting a drafting assistant to your own templates, and switching on time capture inside your billing software are afternoon projects using tools your firm often already owns. Where small firms actually get stuck is deciding which problem to fix first, writing the intake questions and confidentiality guardrails, and getting the tools to talk to your calendar and case files instead of becoming one more place to check.
This is general information about running a law practice, not legal advice, and nothing here replaces your professional-responsibility obligations.
The Free AI Day
One full day inside your business. Free.
I spend a full working day with you and your team, watching how the work actually gets done, and hand you a written plan of the three to five places AI will pay off first, what it costs, and what it does not touch.
Free. No card, no contract, no catch.