- Good first uses: intake, conflicts preparation, document review support and first drafts for attorney review.
- ABA Formal Opinion 512 (July 2024) applies existing duties to AI: competence, confidentiality, communication, candor, supervision and reasonable fees.
- An attorney verifies anything that goes to a client or a court.
- Client information stays inside the firm's access rules and ethical walls.
- Check your state bar's own guidance; ABA opinions are not binding on their own.
The hours nobody bills
A potential client calls at 4:50 on a Thursday. Someone takes notes. Someone else runs the names through conflicts. A paralegal builds the matter. An associate writes the engagement letter.
None of it is billable. All of it has to be right.
That's the best place for AI in most firms: the careful, repetitive work around the legal work, with a lawyer checking anything that matters.
Where AI actually helps
Intake
Web forms, emails and call notes arrive in different shapes. AI can turn each one into a clean intake summary: who, what happened, when, which practice area, and what's missing.
The intake team reviews the summary and decides whether to take the next step. Prospects hear back the same day instead of the next week. Our client onboarding page covers the hand-off from intake to an opened matter.
Conflicts preparation
A conflicts check is only as good as the names that go into it. Intake notes mention the other driver, the landlord's management company and a former spouse. Someone has to find all of them.
AI can pull every party, related company and alternate spelling from the intake material into a list for the conflicts team. The search still runs in your system. The decision still belongs to the person responsible for conflicts.
Document review support
A production lands with 9,000 pages. Before formal review, the team wants to know what's in it.
AI can sort documents by type, summarize each one with page references and flag likely privileged material for a lawyer to look at. It speeds up the first pass. It doesn't replace the review protocol your matter requires.
First drafts
Engagement letters, routine motions, discovery requests, client update emails. AI can produce a first draft from your firm's own templates and the matter facts.
An attorney edits and approves every draft. Citations get checked against the source, every time.
Matter status for clients
Clients call to ask where things stand. AI can draft a plain-language update from the matter's recent activity for the responsible lawyer to approve and send. Fewer calls, better communication.
Deadlines and docket entries
Court notices arrive by email. Someone reads each one, works out what deadline it sets and enters it into the calendar.
AI can read the notice, propose the entries and show the text it relied on. Your docketing staff confirm every date against your rules before it's calendared. A missed deadline is too costly to leave to a draft.
What the ABA says
On July 29, 2024, the ABA Standing Committee on Ethics and Professional Responsibility issued Formal Opinion 512, its first formal opinion on generative AI. It doesn't create new rules. It explains how existing Model Rules apply.
- Competence. Lawyers need a reasonable understanding of what a tool can and can't do. Relying on its output without an appropriate degree of independent verification could fall short of the duty of competence.
- Confidentiality. Before client information goes into a tool that may learn from it, the opinion calls for the client's informed consent. Boilerplate in an engagement letter isn't enough.
- Communication. In some situations, lawyers need to tell clients about their use of these tools.
- Candor and meritorious claims. Filings must be checked so that nothing false reaches a tribunal.
- Supervision. Managing and supervising lawyers should set clear policies and make sure lawyers and staff follow them.
- Fees. Fees must be reasonable. Time spent learning a general tool is generally not billable to the client.
ABA opinions guide, but they don't bind. Many state bars have issued their own guidance. Read your state's before you choose tools or write a policy.
Where it doesn't help (yet)
Legal research you don't verify. Models can invent cases and quotes that look real. Every citation needs a check against the source.
Legal advice without a lawyer. A tool that answers a prospect's legal question directly risks giving advice. Intake tools should gather facts and route, not advise.
Strategy. Which motion to file and when to settle are judgment calls. AI can organize the record. It shouldn't make the call.
Questions to settle before you build
- Which tasks are allowed, which need a partner's approval, and which are off limits?
- How does each tool handle client information under its business terms, and does that trigger a consent conversation?
- Which matters and documents can the system see, and does it respect existing ethical walls?
- Who verifies citations and facts before anything goes to a client or a court?
- How will you handle billing for work that now takes a fraction of the time?
Answer these in a short written policy before choosing a tool. It's far easier than retrofitting rules onto a tool people already use.
How to start safely
Write the policy first. Which tools, for which work, with whose approval. Opinion 512's supervision point makes this a management job, not an IT job.
Pick one job. Intake summaries or conflicts preparation make good starts. The stakes are manageable and the review step already exists.
Keep ethical walls in force. The system should see only the matters the person using it can see. Access comes from your document management rules, not from a new set.
Review everything that leaves. A named lawyer approves every client communication and every filing. Our guide to human review for AI covers how to make that review practical.
Measure it. Hours from first contact to engagement letter. Hours per conflicts check. Write down today's number first.
What a realistic first release looks like
Take intake summaries as the example. The first weeks go to agreeing on what a good summary includes and collecting real intake notes from each practice area.
Then the system drafts summaries alongside the intake team's own work. The team marks what was missing or wrong, and the rules get tightened.
Once the team trusts the drafts, they start from them. After a month or two, you compare time from first contact to decision against your before number.
If it helped, conflicts preparation and engagement letters are natural next steps, since they use the same intake facts.
What changes for your team
Intake staff spend less time typing and more time talking to people. The summary is ready when the call ends.
The conflicts team gets a fuller list of names, faster. They still run the search and make the call.
Associates start from a draft built on the firm's own forms. Their time moves from formatting to judgment, which is the part clients actually value.
Partners carry a new responsibility: making sure the firm's policy is followed. Opinion 512's supervision point makes that explicit. Short training, clear rules and a named owner go a long way.
What most guides miss
Consent depends on the tool. The opinion's informed consent point is tied to tools that may learn from client information. How a given tool handles your data, under its business terms, is the first question to answer. It shapes everything else.
Your templates are the asset. Drafting quality depends on starting from your firm's own forms, not from a general model's idea of a motion. Clean up the template library first.
Billing changes before the tools do. If a draft takes 20 minutes instead of two hours, hourly billing for that work changes. Decide how you'll handle it before clients ask.
Staff adoption beats features. A tool paralegals trust and use daily is worth more than a powerful one nobody opens.
Confidentiality is about people too. Model Rule 1.6 asks lawyers to make reasonable efforts to prevent unauthorized disclosure of client information. Staff pasting a client email into a free consumer tool is the more likely risk than any carefully chosen system. Give people an approved tool and they'll stop improvising.
How we do it
We work alongside your lawyers and staff, starting with a free intro call to find the first job worth fixing. We build on the systems you already run, such as Clio, Filevine or iManage, inside your access rules. See our law firms page and our AI systems work.
