Jump to section
Most intake calls do not need a lawyer on the line for the first ten minutes. They need a clear picture of what the caller is dealing with, whether the firm can help, and what happens next. That picture is what triage produces. Done well, triage protects your time, protects the caller from false hope, and gets the right files to the right people faster. AI can help with parts of that job. It cannot replace the judgment, the warmth, or the professional responsibility that makes intake work.
This post is about the parts where AI genuinely helps, the parts where it should stay out of the way, and the practical steps a small firm can take right now without overcomplicating things.
What Triage Actually Means in a Small Firm
In a busy practice, intake triage is the sorting step between "someone called" and "a lawyer is working the file." It answers four questions: Is this within the firm's practice areas? Is there a limitation period or urgent deadline? Is there a conflict? And what information does the fee earner need before the first real conversation?
Small firms often skip this step, or compress it into the intake call itself, which means the lawyer is doing administrative sorting at $300 or more per hour. That is expensive sorting. It also means the caller gets less of the lawyer's attention on the parts that actually require a lawyer.
A well-designed intake questionnaire, whether delivered by a form, a portal, or a staff member following a script, can answer the first three questions before a lawyer is involved at all. AI can make that questionnaire smarter and can flag patterns in the responses. What it cannot do is assess credibility, read the room, or decide whether an unusual situation might still be a good fit for the firm.
Where AI Adds Real Value in the Triage Step
The most honest use of AI in intake triage is text processing: reading the client's free-text description of their situation and tagging it by matter type, urgency signals, and missing information. A model is good at that. It is fast, it is consistent, and it does not get bored on the fifteenth call of the afternoon.
Here is what that can look like in practice. A prospective client fills out a web form describing their situation in their own words. Before the response goes to a human, a model reads it and produces a short internal note: likely matter type, any urgency flags in the text ("court date next week," "signed an agreement yesterday"), and a list of missing details the intake team should collect. A staff member opens the response, sees the note, and can follow up quickly with the right questions rather than starting from zero.
That is not AI making legal decisions. That is AI doing clerical pre-reading, which frees a person to do the parts that require judgment.
A second honest use is conflict-check assistance. If a client submits their name, the opposing party's name, and a rough description of the matter, an AI-assisted tool can run that against your existing client list and flag potential matches for a human to review. The human still decides whether a conflict exists. The AI saves the step of manually searching the database.
Note. The conflict check is a professional obligation, not a feature. Any AI-assisted tool is a search aid, not the check itself. A lawyer still confirms the result. Always.
Where AI Should Stay Out of the Way
Two areas deserve a clear boundary: assessing legal merit and communicating directly with prospective clients without disclosure.
Assessing merit means deciding whether a claim has a realistic chance of success, whether the facts support a cause of action, whether the damages are worth pursuing. This is legal analysis. It requires judgment about credibility, about how courts in your jurisdiction have treated similar facts, and about factors that a text description will not capture. No current AI tool does this reliably enough to act on without a lawyer's review. Using AI output as a shortcut for that assessment puts clients at risk and puts your licence at risk.
The communication question is subtler. Some intake workflows use AI to draft replies to prospective clients. That can work, but only if the prospective client knows they are interacting with an automated system, and only if the reply is reviewed before it goes out. A message that sounds like it comes from a lawyer, but was written by a model and sent without review, creates a solicitor-client relationship problem in addition to a confidentiality one. Keep a human in the loop on every message that goes to someone who has not yet formally retained the firm.
If you are thinking about how to frame the firm's AI use generally, the post on writing an AI policy your small firm will actually use covers the governance side of exactly these questions.
Watch out. An AI tool that auto-responds to prospective clients with legal-sounding content may inadvertently create a duty of care before any retainer exists. Have a lawyer review the template and the workflow before switching it on.
Building a Triage Process That Keeps the Human Visible
The callers who contact a law firm are often in the worst weeks of their lives. They have been served papers, they have discovered something frightening, or someone they trusted has let them down. The first impression of the firm shapes whether they feel they are in good hands.
That experience is not something AI can produce. What it can do is remove friction from the process so a human can show up sooner, more prepared, and with a clearer picture of what the caller needs. A well-triage'd file means the first real conversation starts with the lawyer already knowing the basics, which means more of that conversation can be about the person rather than the form-filling.
A practical structure for a small firm looks like this. The prospective client fills out a short web form or portal intake with the key facts. A model pre-processes the submission and flags urgency and matter type. A staff member reviews the flag, adds any missing details via a quick follow-up message, and schedules a consultation. The lawyer opens the file to find a one-paragraph summary rather than a wall of raw text. That summary was produced by the model and reviewed by the staff member before it was saved to the file.
Every handoff in that chain has a human. None of the AI steps face the client directly. That is the right architecture for a small firm that wants the efficiency without the risk.
For more on what good intake looks like from the first phone call, the post on the first ten minutes of intake covers the human side of the same process, and spotting limitation periods at intake walks through the urgency-flagging piece in detail.
The One Thing That Has to Stay Human
Triage is ultimately a decision: this firm can help this person, or it cannot, and here is what happens next. That decision carries professional weight, and that weight belongs to a lawyer. AI can inform the decision. It can flag urgency signals, suggest missing information, and remove the clerical steps that slow everything down. It cannot make the call.
The firms that use AI well in intake are not the ones that automated the most steps. They are the ones who were honest about which steps require a person, built the AI around those steps, and made sure the client always knew a human was accountable for the answer. That is not a complicated design. It is just a disciplined one.
If you want to see how these tools fit into the broader picture of running a modern small firm, the legal tech and AI cluster covers everything from vendor selection to safe drafting automation. And if you are ready to manage intake inside a system built for small Canadian firms, A1 CMS is worth a look.
