TL;DR
- Most law firms hiring an AI consultant are trying to fix two things at once: document workflows and timekeeping. Those are the right places to start.
- The hard part is not the technology. It is the change management, the ethics review, and the training that makes attorneys actually use the tools.
- A good engagement starts with an audit of your current stack (Perfect Law, NetDocuments, SharePoint, iManage, whatever you are running) before anyone touches AI.
- Be careful who you hire. Most people calling themselves AI consultants have never sat inside a law firm and watched a partner record time.
A managing partner at PI firm called me last month with a question that is becoming common. The firm had decided it was time to bring AI into the practice. Document review, drafting, billing accuracy, search across their matter history. Leadership was on board. The budget was approved. They just needed someone to actually make it happen.
The question was simple. What does that person do, and how do you tell a real one from someone who watched a few YouTube videos and printed business cards?
This post is the answer.
The two AI problems every firm is trying to solve
When a firm decides to hire an AI consultant, they almost always have the same two pain points driving the decision.
The first is document management. Attorneys are spending hours doing things that should take minutes. Drafting boilerplate that has been drafted a thousand times before. Searching for a specific exhibit across a decade of closed matters. Reviewing discovery productions one document at a time. Reading insurance policies looking for the one clause that matters in this case.
The second is timekeeping and billing. Attorneys hate recording their time, which means a meaningful percentage of billable work never makes it onto an invoice. Block billing creates write-downs. Inconsistent time entries create client complaints and bar complaints. Billing reviewers spend hours editing entries that should have been clean the first time.
These two problems are connected. They are both about capturing the work that has already happened and turning it into something the firm can use. They are also both excellent targets for AI, which is why they tend to be the first things a real consultant tackles.
What the AI audit phase actually looks like
Before anything gets implemented, the engagement starts with an audit. This is the part most firms want to skip, and it is the part that makes the difference between a successful rollout and an expensive disappointment.
The audit covers four things.
Your current document management system. If you are running Perfect Law, the integration path looks one way, NetDocuments or iManage looks different than that, and SharePoint with some custom folder taxonomy looks very different from all of the above. Most firms have some combination of all of the above, plus a shared drive nobody admits to using. The first job is to map what is actually there.
Your timekeeping workflow. How do attorneys currently record time? Are they using a passive capture tool? Reconstructing the day at 6 p.m. from email and calendar? Billing in real time inside a matter management system? Each of those starting points changes what AI can actually help with.
Your compliance posture. Insurance defense work has client confidentiality requirements that vary by carrier. Some carriers have explicit AI policies in their outside counsel guidelines, some are silent and some prohibit it outright. Before any AI tool touches matter data, someone needs to read the engagement letters and the outside counsel guidelines.
Your team’s actual comfort level. A senior partner who still prints emails is going to need a different rollout than an associate who has been using ChatGPT on their own for two years. Most firms have both, sitting in offices next to each other.
If a consultant skips the audit and starts proposing tools in the first meeting, that is the signal to end the conversation.
What gets implemented, in what order
Once the audit is done, a reasonable engagement looks something like this.
Phase one: document workflows
This is usually the highest-leverage starting point because the wins are visible inside thirty days.
The work involves connecting your document management system to AI tools that can read across it. For an insurance defense firm, that often means setting up the ability to ask questions across an entire claim file at once. Summarize this deposition. Pull every reference to the prior accident. Compare the IME report to the treating physician’s records. Draft a letter to the carrier explaining the coverage position.
The technology to do this exists and works. Vendors in this space include Harvey, Spellbook, Casetext (now part of Thomson Reuters), Lexis+ AI, and Claude for Work used carefully with the right data setup. None of them is the right answer for every firm. A consultant who recommends one tool before understanding your stack is not consulting, they are selling.
If you want a deeper view of how this category is shaped right now, the legal AI guide covers the tool landscape in detail.
Phase two: timekeeping and billing
Once document workflows are stable, attention shifts to the time and billing side.
This is where passive time capture comes in. Tools like Ajilon, Smokeball, Time by Ping, and the AI features inside billing platforms like Centerbase and CARET Legal can reconstruct an attorney’s day from email, calendar, document access, and phone activity. The AI does not do the billing. It does the remembering, which is the part attorneys are bad at.
The accuracy and ethics work happens on top of that. AI can flag block billing, suggest task code matches, catch entries that exceed reasonableness thresholds, and surface entries that are likely to be cut by a carrier or auditor. The goal is not to replace attorney judgment. The goal is to make sure attorneys are billing all the time they actually worked, in language that holds up to review.
Done well, this phase typically recovers between 6 and 12 percent of previously lost billable time. For a fifteen-attorney firm with average rates, that math gets very interesting very quickly.
Phase three: training and adoption
Here is the part most firms underestimate.
You can buy the best AI tools in the world and watch them collect dust if the firm does not actually use them. Adoption is a change management problem, not a technology problem. The work involves live workshops by practice group, written guides written for lawyers (not for IT), one-on-one sessions with the partners who are skeptical, and ongoing office hours for the questions that come up after week three.
We have seen firms run this phase well and firms run it badly. The pattern for running it well is consistent. The training is hands-on, not theoretical. It uses the firm’s own matters as examples. It addresses the ethics questions head-on rather than glossing over them. And it accepts that some partners will adopt in week one and some will adopt in month nine, and both timelines are fine.
For more on what AI adoption looks like in practice across different firm types, the Claude AI for lawyers guide goes deeper on the workflow side.
The ethics piece nobody wants to talk about
Any AI consultant worth hiring spends real time on the ethics and confidentiality work, because the rules are not optional and the consequences of getting it wrong are professional, not just operational.
The questions a serious engagement answers, in writing:
- Which AI tools have the right enterprise agreements in place to be used with client data?
- What happens to the data the firm submits? Is it used to train models? Is it retained? For how long?
- What does the firm need to disclose to clients? When? In what form?
- How does the firm handle outside counsel guidelines that prohibit or restrict AI use?
- What is the supervision policy for AI-generated work product? Who reviews what before it goes out?
- How does the firm document its AI use in a way that satisfies ABA Formal Opinion 512 and applicable state bar guidance?
The answers vary by firm, by practice area, and by jurisdiction. A consultant who treats this as a checkbox exercise is going to create problems. A consultant who treats it as foundational is doing the work right.
How to actually evaluate a AI consultant for your law firm before you hire one
If you are reading this because you have been considering an engagement of your own, here are the questions worth asking before you sign anything.
Have you actually worked inside a law firm? Not “worked with law firm clients.” Worked inside the operations. Sat in on partner meetings. Watched intake. Reviewed bills before they went out. The work is different from the outside.
What does your first thirty days look like? If the answer is “we start implementing tools,” walk away. The answer should involve listening, mapping, and writing things down before anything gets touched.
Which document management systems have you actually configured? Perfect Law, NetDocuments, iManage, SharePoint, Filevine, Clio, MyCase. Specific names. If the answer is vague, that is the answer.
Show me a training deck. Not a sales deck. A real training deck used with a real firm. The quality of the training material is the closest proxy to whether the engagement will succeed.
Who else is on your team? AI consulting for a law firm is rarely a solo job. There is usually a technical implementation person, a training person, and someone who runs project management. If it is one person, ask what happens when something breaks at 9 p.m. before a trial.
What does the handoff look like? A good engagement ends with the firm able to operate the systems without ongoing dependency on the consultant. Open-ended retainers that never end are a red flag.
The shorter version when hiring an AI consultant for your law firm
If you are running a firm and you are trying to figure out where to start with AI, the order is almost always the same. Audit the stack you have. Fix document workflows first. Move to timekeeping and billing once that is stable. Build training in parallel from day one. Get the ethics work in writing. Pick tools that fit your existing systems instead of forcing the firm to retool around new ones.
The firms that get this right end up with attorneys who quietly get more done in the same amount of time, billing that captures more of what was actually worked, and partners who stop being nervous every time AI comes up in a conversation with a client.
The firms that get it wrong end up with a tool subscription nobody uses and a partner meeting where someone asks where the budget went.
If you want to talk through what this looks like for your firm specifically, reach out here. We do this work for law firms across personal injury, insurance defense, family law, and a few other practice areas. We are happy to start with a conversation about your current stack before anyone talks about engagement structure.
You can also see how we think about law firm automation more broadly if you want a wider view of what is and is not worth automating in a firm.
