Every employment firm I speak to has the same leak, and it is not billing software, a problem AI for employment lawyers can help address. It is the free fifteen minutes. Someone rings about a personal grievance, you take the call because you are decent and because one in five of them turns into a matter, and forty minutes later you have given away a preliminary view, a timeline, and a rough sense of their prospects. No engagement letter. No file. Nothing on the clock.
It is usually sold as document review or contract drafting. That is the wrong end of the problem. The bigger gain sits in the twenty minutes before the file is even opened: who called, what happened to them, when it happened, and whether you should be acting.
This post covers how to fix intake without sounding like a robot or crossing a regulatory line.
Key Takeaways
- AI for employment lawyers pays off fastest at intake, not in drafting, because intake is where unpaid time and lost enquiries both live.
- A structured intake assistant captures the facts, the dates and the employer name before the first call, so your first conversation is billable-shaped.
- 78% of deals go to whoever responds first, and employment enquiries are especially time-sensitive because the claimant is usually distressed and ringing three firms.
- The 90-day raising period for a personal grievance in New Zealand makes date capture a triage function, not admin.
- An intake assistant qualifies and records. It never gives legal advice, quotes prospects, or predicts outcomes, and it says so out loud.
- Most firms lose more revenue to unreturned after-hours calls than to underpricing, because distressed callers do not leave voicemails.
- Start with one narrow automation, not a firm-wide rollout, and measure the change in enquiries that convert to engagement.
Why the Free Fifteen Minutes Costs More Than You Think
Run the maths on your own week. Say you take six enquiry calls; each one runs twenty-five minutes because employment matters are emotional and people need to tell the whole story before they will answer a single question. That is two and a half hours. If two of those six become files, you have effectively spent an hour and forty minutes unpaid to win two matters.
The cost is not just the time. It is where the time sits. Those calls land mid-afternoon, between drafting and a mediation prep, and they fragment the day in a way that a diarised appointment never does.
Then there is the other half of the leak. The enquiries you miss. An employment enquiry is a distress call, and distressed people do not leave voicemails. They ring the next firm on the search results page. Research on lead response consistently shows the first responder wins around 78% of the time, and that figure does not soften because the caller has a legal problem.
So you have a channel that is expensive when it works and invisible when it fails. That is a systems problem, not a discipline problem. You are not going to fix it by being more disciplined about scoping the free call.
What an Intake Assistant Actually Does
The version that works in a small employment practice is narrow and unglamorous. It answers, it asks, it records, it routes. That is the whole job.
It answers every enquiry, including after hours
Voice AI picks up when reception is on lunch, when you are in mediation, and at 8 pm when someone has just been handed a letter about a disciplinary meeting on Monday. It says who it is, it says it is an assistant rather than a lawyer, and it collects the basics.
I have seen this change the numbers in other appointment-driven practices. One dental client, Dr Claire, went from a significant share of calls unanswered to zero missed calls and 44% more booked appointments. The mechanics are identical in a law firm: the phone rings, someone competent picks up, the caller gets a next step instead of a dial tone.
It captures the facts that determine triage
For employment work, the triage questions are specific, and they never change. Was the person dismissed, or are they still employed? What date did the thing they are upset about happen? Have they raised anything in writing? Is there an employment agreement, and do they have a copy? How many people does the employer have? Was there a support person at the meeting?
That last set matters because of the 90-day rule. In New Zealand, a personal grievance normally has to be raised with the employer within 90 days of the event, and Employment New Zealand sets out the limited exceptions. If an assistant captures the date of dismissal on the first contact, you know immediately whether an enquiry is urgent, borderline, or already difficult. That is a triage decision you are currently making in the nineteenth minute of an unpaid call.
It writes the file note before you read it
Every enquiry lands in your system as a structured summary: name, employer, dates, what happened in their words, what they want, and a flag for anything time-critical. You open it in ninety seconds and decide whether to call, decline, or refer.
The Line You Cannot Cross
This is the part that stops most lawyers from trying, and rightly so.
An intake assistant must not give legal advice. Not a preliminary view, not “that sounds like an unjustified dismissal”, not a range of likely outcomes. It collects information and books time. If a caller pushes for a view, the correct behaviour is a short, honest deflection: the assistant is not a lawyer, cannot advise, and the lawyer will address it on the call.
Three rules I would build into any employment law intake system:
Disclose that it is AI, immediately. Not buried, not on request. Within the first sentence. The Rules of Conduct and Client Care do not let you mislead anyone about who they are dealing with, and a caller in distress deserves to know before they start talking.
No quoting, no prospects, no timeframes. Fees and merits are lawyer territory. The assistant offers times, not numbers.
A human reviews before engagement. Nothing becomes a file, and no conflict check is cleared, without you looking at it. The system prepares the decision. It does not make it.
Handled that way, an intake assistant is closer to a very consistent legal secretary than to anything experimental. It just never gets tired at 6 pm and never forgets to ask the dismissal date.
Where AI for Employment Lawyers Fits Beyond Intake
Once the front door works, the same underlying idea applies across the practice. Each of these is a separate, narrow automation, not one giant system.
The dormant enquiry list. Every employment practice has a list of people who rang, got a quote or a preliminary chat, and went quiet. Some settled. Some did nothing and are now in a worse position. A structured reactivation campaign re-opens those conversations. In one finance business we ran this for, 319 dormant contacts the team had written off produced $49,000 in recovered revenue. The principle carries: people who already contacted you are the warmest audience you own.
Deadline monitoring. The 90-day raising period, the three-year limitation for most employment claims, mediation dates, Authority filing dates. A system that reads your matter data and surfaces what is approaching this week is worth more than any drafting tool.
Document intake, not document drafting. Clients send employment agreements, warning letters, text message screenshots and email chains in whatever format they have. Having those read, named, dated and filed against the matter before you open them saves real time and removes a genuinely awful job from a junior’s week.
The morning brief. New enquiries since yesterday, matters with something due this week, anything a client escalated. Five minutes of reading instead of forty minutes of checking.
What connects all of these is not the tools. It is that the business has somewhere to keep what it knows. Most small firms run on the principal’s memory plus a practice management system that stores documents but does not think. The shift is giving the practice a brain that holds the context, and a workforce that acts on it.
How to Start Without Betting the Practice
Pick the narrowest version of the problem and fix only that.
For most employment practices, that is after-hours and overflow call handling. It is the leak with the clearest revenue attached and the lowest risk, because the assistant is doing something nobody is currently doing at all. You are not replacing a person. You are covering the hours when the phone rings out.
Run it for thirty days and measure three things. How many enquiries were captured outside office hours. How many of those converted to a first appointment. How much of your own unpaid enquiry time disappeared.
If those numbers move, add the next thing. Structured intake during business hours, then dormant enquiry reactivation, then deadline monitoring. One at a time, each one proving itself before the next arrives. You will learn more from one working automation than from six months of evaluating platforms.
The firms that get stuck are the ones that try to design the whole thing first. There is a useful parallel in our post on admin overload in small business: the trap is always scoping the perfect system instead of removing the single worst task. And if you have been burned before, what to do after a bad vendor covers how to restart without repeating the mistake.
The Real Problem Behind the Free Fifteen Minutes
For firms exploring AI for employment lawyers, the free fifteen minutes is not a pricing failure. It is a symptom of a practice where every judgement call, every piece of context and every decision about whether a matter is worth taking routes through one head. Yours.
That is why the calls have to be yours. That is why enquiries go cold when you are in mediation. That is why the practice cannot grow past the number of hours you can personally sit at a desk, and why stepping away for two weeks means coming back to a fortnight of backlog.
Fixing intake buys back a few hours a week, which is worth doing. But the bigger prize is a practice that holds its own knowledge: who the clients are, what stage each matter is at, what has to happen this week, and what a new enquiry needs before it reaches you. Once that exists, the automations stop being separate tools and start compounding.
You built a legal practice, not a switchboard. It should not need you at the switchboard to work.
Get a Straight Answer on What to Fix First
If you want to know which part of your intake is actually costing you money, book a 30-minute discovery call. No pitch deck. We go through how enquiries reach you now, where they leak, and what the first automation should be. If the answer is that you do not need one yet, I will tell you that.
Frequently Asked Questions
Can AI give legal advice to clients?
No, and it should not try. An intake assistant collects facts, books appointments and writes file notes. It does not assess merits, predict outcomes or answer legal questions. If a caller asks for a view, the correct response is that the assistant is not a lawyer and the lawyer will cover it on the call. Every enquiry gets reviewed by a person before anything becomes a matter.
Do I have to tell callers they are talking to AI?
Yes. Disclose it in the first sentence, not on request. Beyond the professional conduct obligation not to mislead anyone about who they are dealing with, it is simply better practice. Callers accept an AI assistant readily when it is upfront and competent. They react badly when they work it out halfway through and feel they were tricked.
What does AI for employment lawyers actually cost to run?
Less than a part-time reception hire, and the setup is usually a low four-figure investment depending on scope. The more useful comparison is the cost of missed enquiries. If your practice loses two matters a month to unanswered after-hours calls, the system pays for itself several times over before you count the unpaid enquiry time you get back.
Is it safe to put client information into an AI system?
It depends entirely on the setup. Data should sit in systems you control, under a standard confidentiality and privacy arrangement, not pasted into a consumer chatbot. Ask any provider where the data is stored, who can access it, and whether it is used to train models. If they cannot answer those three questions clearly, do not proceed.
Will clients find an AI receptionist off-putting for sensitive matters?
Less than you would expect, provided it is honest, and the handover is quick. The comparison is not a warm human conversation. It is a phone ringing out at 7 pm while someone sits with a dismissal letter. A competent assistant that takes details and confirms a callback time is a far better experience than voicemail or silence.
How long does it take to set up intake automation for a law firm?
A narrow first build, after-hours call handling with structured capture, typically runs a couple of weeks from scoping to live, most of which is writing the qualification questions and testing them. Broader work takes longer. The thing that slows firms down is not the build; it is deciding what the assistant should and should not say.
What should I automate first in a small law firm?
Whatever is losing you the most money while requiring the least judgement. For most employment practices, that is enquiry capture outside office hours, because nobody is currently doing it and the revenue attached is easy to measure. Avoid starting with anything that needs legal judgement. Those are the last things to touch, not the first.
About Octavius
Titus Mulquiney is the founder of Octavius AI, where he builds AI brains and AI workforces for founder-led businesses stuck running everything out of their own head. Twenty years in marketing, ex-Sony product manager, ex-GM Zeal NZ. Based in Auckland, working with operators across NZ, Australia, and the US. Connect on LinkedIn.