"What should I say when the lawyer's office calls me back?" That's the kind of question a prospective client is now typing into a chatbot at 11 p.m., before the firm's intake coordinator has ever heard their name. By the time that person fills out the contact form the next morning, they arrive with a printed summary of their situation, a rough theory of liability, a list of questions, and sometimes a draft of their own facts written in the voice of a paralegal.
On the other side of that form, a different change is underway. Associates are drafting with generative tools, paralegals are using AI to summarize discovery, and partners are asking hard questions about what can be billed, what has to be supervised, and what the ethics rules now require. The intake desk is where those two changes meet, and where firms that have treated them as separate initiatives start to feel the friction.
The Prospect Arrives Already Coached
The person on the other end of the intake call isn't the same person who called five years ago. They've already described their situation to a chatbot, been handed a plain-language explanation of the relevant doctrine, and been told what documents to gather. A large study of how the public uses AI chatbots for legal help found that most people still combine the chatbot with a human source, but a meaningful share are treating the AI as their first and sometimes only advisor before deciding whether to call a firm at all.
That changes what the intake desk is actually doing. It's no longer capturing a raw problem. The coordinator is validating, correcting, or gently redirecting a story the prospect has already half-built with a machine. If the person answering the phone handles the call the way they did in 2019, the firm loses the file to someone who sounds like they were expecting the coached version.
The Same Desk Is Where the Firm's AI Shows Up First
Inside the firm, AI doesn't usually debut in the courtroom. It shows up at intake, in the form of conflict checks, form parsing, call summaries, and first-pass matter classification. That's convenient, until you notice the prospect and the firm are now running AI on opposite sides of the same conversation, with no shared understanding of what either system is doing with the information.
This is the piece most firms miss when they scope their internal AI project and their client-acquisition project as separate workstreams. They are, in practice, the two AI disruptions hitting law firms at the same time, and they collide at the exact point where a prospect becomes a client. Treating them as one program, with one owner, one policy, and one measurement layer, is what stops the collision from turning into lost matters and avoidable risk.
The Collision Creates Risks Nobody Was Watching For
When an AI-coached prospect meets an AI-assisted intake desk, new problems appear that neither side planned for. They're the boring, procedural things that decide whether a matter is worth taking and whether the firm can defend its work later.
- Privilege exposure before the engagement letter. A prospect who pasted their facts into a consumer chatbot may have already weakened their own position. A National Law Review analysis walks through recent rulings that treat AI-generated material as discoverable. Intake needs to ask about it.
- Fabricated confidence. The chatbot gave the prospect a clean answer. The lawyer's honest answer is messier. If intake doesn't explain why, the firm looks evasive on the first call.
- Billing questions on day one. Prospects now ask whether AI will be used on their matter and how it affects the bill. The firm needs a consistent answer ready before it's asked.
- Speed expectations. A person who got a response in three seconds from a chatbot is unforgiving about a two-day callback.
One Program Beats Two Pilots
The firms getting further are the ones that stopped funding two pilots and started running one program. In practice, that means a single policy on what AI can touch, one person accountable for how prospects are handled from first click to signed engagement, and one measurement layer that watches the whole path rather than the marketing funnel and the matter workflow separately.
It also means writing down the answers to questions the ethics rules already care about. ABA Formal Opinion 512 is the current reference point on generative AI, and a useful walkthrough from UNC Law lays out the six duties partners should map their tools against: competence, confidentiality, communication, candor, supervision, and reasonable fees. Those duties apply to the chatbot on the website and the intake script on the phone as much as they apply to the associate drafting a brief.
The intake desk is the smallest surface in the firm where all of this shows up at once. Watch what happens there for a month and you can tell whether the firm is running one program or two. The prospects will tell you first.