Can a NJ Solo Attorney Actually Use AI for Conflicts Screening? What the Ethics Rules Require

Conflicts of interest checks have always been one of those tasks that solo attorneys handle informally until the day they can't. A quick scan of a mental Rolodex, a search through an old email folder, maybe a glance at a spreadsheet that hasn't been updated in six months. It works until it doesn't, and when it doesn't, the consequences range from a disqualification motion to a grievance filed with the Office of Attorney Ethics.

AI-assisted conflicts screening is being marketed hard right now as the fix. Several practice management platforms, including Clio, MyCase, and a handful of standalone tools, have built or are building AI-powered intake and conflicts modules. The pitch is compelling: the system ingests your past client data, flags name matches and relationship overlaps, and surfaces potential conflicts before you open a new matter. For a solo practitioner with five years of files and no dedicated staff, that sounds like exactly what the job needs.

But before you hand this task to any automated system, you need to understand what NJ's ethics rules actually demand, because the rules set a floor that no vendor will set for you.

What RPC 1.7 and 1.9 Actually Require

RPC 1.7 covers concurrent conflicts, and RPC 1.9 covers duties to former clients. Neither rule says anything about how you conduct a conflicts check. What they do say is that you cannot represent a client if a concurrent or successive conflict exists, and that ignorance of a conflict is not a defense.

The New Jersey Supreme Court has consistently interpreted competence under RPC 1.1 to include the administrative systems a lawyer uses to run a practice. The ACPE has signaled in prior opinions that a lawyer's office systems, including intake and conflicts procedures, are part of the competent management of a firm. That means if your AI-assisted screening tool misses a conflict because you never loaded your prior client data into it properly, or because you accepted its output without verification, the ethical exposure sits with you.

This is the gap most vendors won't close for you.

The Data Problem Nobody Talks About in the Sales Demo

An AI conflicts tool is only as good as the data you feed it. For a solo practitioner who has been practicing for more than a few years, that data lives in a lot of places: a legacy case management system, an Outlook contacts folder, a filing cabinet of paper retainers, a Google Workspace account from a prior firm, billing records in QuickBooks. None of those sources automatically talk to each other.

Before any AI screening tool can be trusted, you need a complete, deduplicated client and matter list that the tool can actually search against. That means past clients by full legal name and any aliases you recorded, adverse parties from prior matters, and related entities where you represented one party in a corporate or transactional context. If you haven't built that list, the AI has nothing useful to screen against.

This is a data governance project before it's an AI project. It probably takes a solo attorney two to four hours to pull together if records are reasonably organized, and longer if they aren't. Some attorneys find the process surfaces conflicts they had forgotten about entirely.

What Verification Has to Look Like

Once a screening tool runs a check, the output is a starting point, not a conclusion. AI name-matching is genuinely useful for catching the obvious, but it misses phonetic variations, name changes after marriage or divorce, entities that share a common principal, and opposing-party relationships that were never tagged at intake. A system that confidently returns "no conflicts found" can still be wrong if the underlying data is incomplete or the query was too narrow.

A practical verification step for NJ solos: after running the automated check, personally review the new client's stated adverse party against your matter list using a second search method, even just a text search in your email client. It takes three minutes. That cross-check is what converts an AI output into a defensible, attorney-reviewed conclusion.

Document the process. A short note in the new matter file, something like "conflicts check run on [date] via [tool], cross-verified by attorney against prior matter list, no conflicts identified," creates a record that you exercised professional judgment rather than delegated it wholesale.

Practical Setup Steps Before You Trust Any AI Conflicts Tool

Start with your data before you start with the software. Export your full prior client and matter list from whatever system holds it, clean it for duplicates, add known adverse parties where you have them, and save that as a running master file you update at intake for every new matter. Then evaluate whether any conflicts module you're considering actually ingests and searches that file format, or whether it only screens against matters entered natively into that platform going forward.

Ask the vendor one specific question: does the tool flag partial name matches and phonetic variants, or only exact strings? The answer tells you how much manual judgment the tool requires at the back end.

For NJ solos who take on both transactional and litigation work, also confirm whether the tool tracks adverse party relationships, not just client names. A conflicts system that only knows who your clients were is incomplete for anyone who does litigation.

The technology here is genuinely useful. But conflicts screening is a legal judgment call, not a data retrieval task. The attorney who uses an AI tool to surface candidates and then makes the call is doing the job right. The one who treats the tool's output as the answer has outsourced something the RPCs say can't be outsourced.