Every agency now claims to ‘do AI’. Almost none understand a law firm well enough for it to matter. Most of the noise about AI in legal marketing is made by people who learned the word last year. We have worked with law firms, and nothing else, for more than two decades. That is why we are careful about what AI can and cannot do for a practice, and honest about where it has no business being.
It is tempting to treat AI as something you buy, a chatbot for the website, a drafting tool for the lawyers, a subscription that arrives and quietly makes the firm modern. That framing is comfortable, and it is wrong. A tool installed on top of a process nobody has examined does not improve the process. It accelerates it, including all the parts that were already broken.
The useful question is not “which AI tool should we get.” It is “where, specifically, does work move too slowly, drop between people, or depend on someone remembering, and what would it be worth to fix that.” Answering it requires understanding how a particular firm actually operates: how an enquiry arrives and what happens to it in the next ten minutes; how a matter opens; when and how a conflict is checked; which tasks consume senior time that could be spent on work only a senior person can do. That is the diagnosis. The technology is what comes after.
This is the distinction a generalist agency cannot make, because they have never sat inside the process. They can install a chatbot. They cannot tell you whether a chatbot is the right answer for your firm, what it must never say, or how it should hand a genuine enquiry to a human before the prospective client loses patience. We can, because for twenty years it is the only kind of business we have worked in, and because we build on software we wrote specifically for law firms, rather than generic platforms bent to approximate one.
Consider a personal injury firm. Its marketing is working, enquiries arrive through the website, by phone, and after hours. But each enquiry then enters a manual process: someone has to notice it, judge it, and act before it goes cold. During a busy week, some are handled well, some sit unread, and a few get a callback two days later, by which point the prospective client has already retained the firm that answered first. The firm cannot see this happening, because a lead that was never logged leaves no trace. The instinct is to spend more on marketing. The actual problem is everything that happens after the click. That is a diagnosis problem, not a tool problem, and it is the one most firms get backwards.
“Our firm has benefitted enormously from the expertise that Practice Proof brings with marketing. Practice Proof has been second to none in our experience going back many years and we’ve tried plenty of other agencies.”
This is the page most exposed to the ‘everyone does AI now’ problem, so it is worth being precise about exactly where the generic approach breaks.
AI applied to an unexamined workflow simply produces bad output faster. If intake is disorganised, an intake bot makes it disorganised at scale. Most implementations skip the diagnosis entirely because diagnosis is slow and unglamorous and does not demo well, so they automate the mess and call it transformation.
Conflicts, confidentiality, privilege, the Solicitors’ Conduct Rules, the specific shape of a matter, these constrain what can be automated and how. An AI agent that captures intake without flagging a potential conflict, or that says something to a prospective client a solicitor never could, is not a convenience; it is a liability.
A generalist builder does not know these constraints exist until one of them causes a problem. We design around them from the first conversation, because we have spent twenty years living inside them.
Off-the-shelf AI products are built for everyone, which means they are built for no one in particular. They approximate a law firm’s reality, and the fit is never quite right, so the firm ends up bending its process to suit the tool, which is exactly backwards. Our systems run on a stack we built for legal practice, so they fit the firm rather than the other way round.
The genuinely dangerous version of legal AI is the one trusted to exercise judgment it does not have. AI does not do the thinking, and it should never be the last word on anything that requires a lawyer’s judgment. Used properly, it removes the manual, repetitive, easily-dropped work, so the people who do think — your senior lawyers — have the time and attention to do more of it. That is the whole point, and it is the opposite of replacement.
We map how your firm operates, find where automation recovers the most value, and, just as usefully, where it will not, then build systems that fit, in the order that compounds.
We trace the real path of work: how enquiries arrive across every channel, what happens in the minutes after, how a matter opens, when conflicts are checked, and which tasks quietly consume senior time. This is the step generic implementations skip, and it is the one that makes everything after it worth doing.
We identify where automation recovers the most value for your firm specifically, and where it recovers little and should be left alone. A recommendation to not automate something is as valuable as the reverse, and you will get both.
AI surfaces the patterns across your data; the people who understand law set the prioritiesFirm-trained AI workspaces, intake and qualification automation, workflow automation that removes manual handoffs, and the AI search-visibility layer, built on our own legal stack rather than generic plugins, and shaped to the diagnosis rather than to a product roadmap.
Your lawyers set the judgment and the compliance line; AI gives the speed and the scaleWe run the systems for you, or we document them, hand them over, and train your team to operate them, with ongoing support whenever you need us. Same engineering, same legal understanding, the only question is who holds the controls day to day.
Take a family law practice with a capable office manager but no marketing or technical staff. After the diagnosis, the highest-value fix is not a grand AI overhaul, it is an after-hours intake system that captures the facts of a separation enquiry, runs a preliminary conflict check, and routes it to the right solicitor for the morning, with a draft engagement ready. Built around how that firm actually opens a matter, it recovers enquiries that previously reached voicemail and were lost. Handed over and documented, the office manager runs it. The work that needed a senior person, the judgment, the first real conversation, still goes to one. That is AI as accelerant, not replacement.
It is easy to treat this as something to get to next year. The difficulty is that the cost of waiting is invisible, which is exactly why it is dangerous. A firm does not receive an invoice for the enquiries it never logged, the consultations that were booked too slowly to convert, or the senior hours spent on work a system could have handled. The loss is real, but it never appears on a statement, so it is easy to believe it is not happening.
Meanwhile the gap widens. The firm two suburbs away that captures after-hours enquiries, qualifies them in minutes, and frees its lawyers to do senior work is not merely moving faster, it is quietly taking the matters yours is losing to voicemail and slow follow-up. Each month that continues, the advantage compounds in their favour. This is not a reason to panic, and it is not a reason to buy the first tool that demos well. It is a reason to find out, precisely, where your firm is leaking, which is what the audit is for, and why it carries no commitment beyond itself.
No, and if that is all someone is offering you, decline it. A chatbot is one possible output of a diagnosis, not the starting point. The work is understanding your intake, your conflicts process, and your matter flow, then building what fits, which may be a qualification agent, an after-hours voice intake, a firm-trained workspace, or an automation that simply removes a manual handoff. The technology follows the problem, never the other way round.
Most firms we speak to have, and usually for the same reason: the agency was a generalist who treated a law firm like any other small business, billed by the hour, and optimised channels in isolation from whether any of it produced matters. We work with law firms only, on a flat fee, with direct access to Dan, and we diagnose before we build, so the work is grounded in your firm rather than a template. If we are not the right fit, we will tell you that too.
That is precisely what the two-door model is for. If no one internally will own these systems, we run them for you end to end, that is Door 1, and it is the responsible recommendation for most firms without spare operational capacity. Door 2, where we build and hand over, is for firms that do have someone to own it. We will tell you honestly which one fits, rather than selling you the one that needs you to find time you do not have.
“It’s rare to see an agency prepared to invest in external consultants to review its own work. In professional-services marketing these guys are exceptional, and they treat their clients’ money as if it were their own.”
The work compounds when channels connect. These pair naturally with what you just read.
AI intake and qualification only matter if enquiries are arriving. Demand generation fills the top of the funnel the AI then handles.
Explore →The same automation thinking applies after the matter: review velocity, monitoring and compliant responses, handled systematically through FirmRanker.
Explore →Start with a Free Growth Audit. We map how your firm operates, show you the mechanism behind each gap, and give you a straight view of which door fits, before you commit to anything beyond the audit itself.
Get a Free Growth Audit →