Most of the AI conversations happening in family law right now are stuck at the same level.
Someone hears about a new tool, tries it out for a few weeks, gets some incremental value out of it, and moves on.
That’s fine as a starting point.
The firms that are actually pulling ahead have moved past that phase entirely and started thinking about AI as infrastructure.
Adoption in family law is still relatively low.
Our team’s read is that we are somewhere around 20% of firms actively using AI in a meaningful way.
The firms in that group are seeing productivity gains, revenue increases, and a widening gap between themselves and the firms that have not started yet.
This is the industrial revolution moment for our profession. The firms watching from the sidelines will feel the way factory owners felt when the assembly line arrived.
Get the Basics Right First
Before any of the more advanced conversation is worth having, there is a basic issue our team flagged that every firm owner should understand.
There are two types of accounts with the major AI providers:
- Consumer-grade
- Enterprise-grade
The terms of service on those are completely different, and using the wrong one can put your firm in a position where you are violating ethics rules around client data.
Consumer accounts, including most of the free and low-tier paid versions people are familiar with, generally allow the AI provider to use your data to train their models.
If you are entering client information into that kind of account, you have a problem. Enterprise accounts are structured differently.
The data stays yours, and the provider cannot use it for training.
If your firm is going to use AI at all, the first move is making sure everyone on the team is working inside an enterprise account. That single decision unlocks the rest of what is possible.
The Ladder From Prompting to Factory
Once the foundation is set, our team described a progression that is worth understanding because most firms are still on the first rung and do not realize how much further they could go.
The first level is prompting. You open a chat window, ask a question, and get a response. Useful, but roughly 5-10% of what the technology can do.
Our team compared it to driving a sports car and never leaving first gear.
The next level is skills. A skill is essentially a standard operating procedure written for the AI. If you can teach a human how to do something, you can teach the AI how to do it too.
When this happens, that scenario triggers that response. It’s basic but powerful.
Above that, you start connecting skills to your actual data — your case management system, your inbox, your document repository. Now the AI can do useful work with real information rather than generic answers.
The top of the ladder is what our team calls a factory. Multiple skills chained together, running automatically, producing complete outputs without a human triggering each step. One prompt kicks off a whole sequence, and something meaningful comes out the other end.
What This Looks Like in Practice
Our team shared an example that made this concrete for me.
Sterling used to have a quality assurance team that listened to a sample of intake calls each week and scored them against a rubric. Two full-time offshore staff, ten inbound and ten outbound calls per intake agent per week, with results delivered through spreadsheets.
That entire process has been rebuilt as an AI factory.
Every call ends and gets transcribed. Every transcription gets scored against a rubric with dozens of criteria.
Every morning, a report lands with performance data across every call from the previous seven days for every agent.
The coaching happens at 8 a.m. with real data from real calls.
The cost dropped from roughly $2,500 per QA person per month to around $600 per month for the whole system, and the quality of coaching improved because there is more data available than any team could have manually produced.
The people who used to do that QA work moved into more valuable roles inside the firm. That is the framing our team keeps coming back to. AI is not replacing people.
It is taking the laborious, low-value work off their plates so they can spend more time on the parts of the job that actually require a human.
Why This Matters Sooner Than It Seems
The gap between firms that are building this kind of infrastructure and firms that are still just experimenting with chat prompts is going to keep widening. Recent data has AI-adopting firms growing at roughly four times the rate of firms that have not adopted. And that gap compounds over time.
Client expectations are shifting, too. As AI shows up in more parts of daily life, clients are going to start asking why certain tasks took as long as they did or why the bill looks the way it does.
Firms operating on hourly billing models are especially exposed here. Firms operating on fixed fee or flat fee models have more room to absorb the shift because the incentive already runs toward efficiency.
The point is not that anyone needs to build a factory tomorrow. The point is that the direction of travel is clear, and the firms that start climbing the ladder now will be in a very different position two years from now than the firms still sitting at the first rung.
