Five Things NJ Solo Attorneys Should Check Before Using AI for Client Intake

Solo practitioners are under constant pressure to do more with less, and AI-driven intake tools have become one of the most aggressively marketed solutions in legal tech. Chatbots, automated questionnaires, and AI phone screeners can collect matter details, run preliminary conflict checks, and schedule consultations without a human touching the keyboard. The pitch is compelling. The risk is real, and specific.

New Jersey's ethics framework doesn't have a dedicated AI intake rule, but it does have rules that govern every step of the intake process, and those rules don't pause because the entity asking the questions is software. Before you flip the switch on any automated intake workflow, there are five concrete things worth confirming.

---

1. The tool isn't giving legal advice without a license.

This is the threshold issue. NJ RPC 5.5(b) prohibits a non-lawyer from practicing law in New Jersey. An AI system that tells a prospective client "based on your answers, you have a strong premises liability claim" has crossed a line your firm can't walk back. Intake tools should gather facts, not analyze them. Review every response template the tool generates. If it characterizes legal rights, recommends a course of action, or interprets a statute in response to a user's facts, that's unauthorized practice territory, and you're the licensed attorney whose name is on the firm.

2. Prospective client communications trigger confidentiality duties immediately.

Under NJ RPC 1.18, a person who discusses the possibility of forming a client-lawyer relationship is a prospective client, even if you never actually take the matter. Information they share is protected. That means the moment your intake bot starts collecting facts about a legal problem, confidentiality obligations attach to that data under NJ RPC 1.6. Before any AI tool touches those conversations, confirm: Where is that data stored? Who can access it? Does the vendor's terms of service claim any right to use conversation content for model training? A vendor that trains on client-identifying intake data is a confidentiality problem, not just a privacy nuisance.

3. Automated conflict screening at intake has real limitations.

Some intake platforms advertise built-in conflict checking. It's a useful feature in theory. In practice, automated conflict tools only catch what's in their database, and a solo firm's conflict database is only as good as what was entered manually over the years. NJ RPC 1.7 governs concurrent conflicts and NJ RPC 1.9 governs duties to former clients. Neither rule cares whether the conflict check was done by a human or a machine. If the AI intake tool clears a matter and you later discover a conflict the tool missed because of a data-entry gap, you own that problem. Automated screening can supplement human review; it can't replace it.

4. The NJ Supreme Court's AI guidelines apply to intake, too.

The New Jersey Supreme Court issued its "Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence" on January 24, 2024. Those guidelines are not limited to brief-writing or courtroom work. They apply to attorney conduct broadly and make clear that lawyers remain responsible for work product and client interactions regardless of what tool assisted in producing them. If an intake bot mischaracterizes a statute of limitations or misstates the cost of representation to a prospective client, the guidelines reinforce that the supervising attorney bears responsibility for that output.

5. The intake interface itself can create implied attorney-client relationships.

This one is procedural, but it bites. New Jersey courts have found implied attorney-client relationships based on reasonable client expectations, not just formal engagement letters. If your AI intake chatbot is embedded on your website under a banner that reads "Talk to Our Legal Team" and then collects detailed facts about a sensitive matter, a prospective client may reasonably believe they've begun a legal relationship. That expectation matters for confidentiality, conflict purposes, and potential malpractice exposure. The intake interface should make clear, in plain language before any facts are collected, that completing the intake form does not create an attorney-client relationship. That disclaimer needs to be prominent, not buried in fine print at page bottom.

---

What the rule requires

NJ RPC 5.5(b) prohibits assisting a non-lawyer entity in the unauthorized practice of law. NJ RPC 1.6 requires lawyers to take reasonable measures to prevent the unauthorized disclosure of information relating to a client's representation. NJ RPC 1.18 extends confidentiality protections to prospective clients. NJ RPC 1.7 and 1.9 require conflict screening before undertaking representation. The New Jersey Supreme Court's Preliminary Guidelines on New Jersey Lawyers' Use of Artificial Intelligence (January 24, 2024) affirm that attorney responsibility for AI-assisted work is non-delegable.

What I recommend

Treat AI intake tools the way you'd treat a new paralegal on their first week: don't let them run unsupervised. Audit every message template and response script for language that characterizes legal rights or recommends legal action, and strip it out. Add a clear pre-intake disclaimer that no attorney-client relationship is formed until both parties execute a written engagement agreement. Store intake conversation logs in your practice management system, not just the vendor's cloud, so you control the record. Run a parallel manual conflict check for every intake that progresses past the initial screening, regardless of what the automated tool reports. And read your AI vendor's data processing terms before you go live, specifically looking for any right to use conversation data for training purposes. If you find such a clause and can't negotiate it out, that vendor isn't the right fit for intake work.