Stroll the expo flooring at any authorized tech convention proper now and rely the cubicles with “AI” within the identify or the tagline. At ABA TECHSHOW 2026, it was most of them. AI consumption. AI timekeeping. AI drafting. AI for every thing.
Two years in the past, that density would have felt thrilling. As we speak it looks like one thing else—not cynicism precisely, however a form of wariness that wasn’t there earlier than. The attorneys strolling these aisles have modified. They’ve spent a 12 months utilizing Claude, Copilot, ChatGPT. They’ve discovered what AI does simply. And that have has quietly shifted what “AI-powered” means as a promoting level.
It used to open doorways. Now it raises a query: in comparison with what?
The Check
Right here’s the query. Open Claude. Describe the issue the product claims to resolve. See how far you get.
For those who can replicate most of what a product does in a single dialog—free of charge, proper now, with out a subscription—that product isn’t fixing your downside. It’s charging you for a immediate.
That’s the Claude Check. It’s not a knock on AI. It’s not a purpose to cancel your complete tech stack. It’s a filter—a primary query to ask before you purchase, and an sincere query to ask about what you’re already paying for. It separates merchandise constructed on actual foundations from merchandise that borrowed AI’s credibility with out constructing something beneath it.
Subscribe
Get skilled insights and sensible suggestions delivered to your inbox each week.
Why This Check Didn’t Exist Two Years In the past
Constructing an AI-powered authorized product was once technically arduous. That problem created a pure filter. If a vendor had shipped one thing, they’d no less than solved an engineering downside. That filter is gone.
As we speak anybody can construct a wrapper. The price of delivery a product that summarizes paperwork, drafts consumption emails, or generates kind language is near zero. The instruments to construct it are the identical instruments your agency already makes use of. There’s a complete motion—vibe coding—constructed across the thought that you could describe what you need and have working software program in a day.
When constructing will get that straightforward, the merchandise with actual foundations and the merchandise with nothing beneath them begin to look an identical from the surface. Similar touchdown web page. Similar demo. Similar pitch. The Claude Check is the way you inform them aside—as a result of those with nothing beneath will fail it instantly.
What Really Passes
The merchandise that survive the Claude Check share one thing in widespread: their worth isn’t within the output they produce. It’s in what they’ve constructed, gathered, or turn out to be over time. There are six classes value realizing.
Issues that had been already arduous. If fixing one thing required years of labor earlier than AI existed, AI didn’t make it straightforward. It simply made every thing else straightforward by comparability. The dominant doc administration platforms in authorized—NetDocuments, iManage—aren’t useful as a result of they retailer recordsdata. They’re useful as a result of they’ve spent a long time constructing the permissioning structure, compliance infrastructure, safety certifications, and migration tooling that enterprise authorized work requires. These issues took a very long time to construct they usually’d take a very long time to switch. AI didn’t change that. It simply made everybody else’s moat look shallower.
Networks and relationships. You’ll be able to’t immediate a community into existence. ABC Authorized has over 8,000 course of servers nationwide—that’s not a characteristic, it’s a logistics operation constructed over a long time. BriefPoint has dwell manufacturing integrations with Clio, 8am MyCase, and Smokeball. LawToolBox has spent years sustaining court docket guidelines databases that energy deadline calculations throughout jurisdictions. This stuff exist due to relationships and commitments made lengthy earlier than any of them had an AI layer. The AI layer, when it got here, was only a new floor on an actual basis.
Accountability. When one thing goes fallacious, who solutions for it? A agency that makes use of Claude to draft a discovery response owns all of that threat—each phrase, each quotation, each omission. A product that maintains jurisdiction-specific compliance guidelines, carries SOC 2 certification, and produces an auditable document of what it did has shifted a few of that accountability away from the agency. That switch of duty is an actual factor of worth. It isn’t one thing you may ship in model one, and it isn’t one thing a general-purpose AI device supplies. Merchandise that carry accountability—genuinely, contractually, operationally—are offering one thing structural that the Claude Check can not replicate.
Workflow place. There’s a distinction between a product that touches your workflow and one that’s your workflow. The merchandise that turn out to be load-bearing—that your billing connects via, that your DMS indexes towards, that your battle checks pull from—should not replaceable by a greater immediate. Eradicating them is a mission, not a call. That depth isn’t lock-in as a flaw. It’s integration as a characteristic. The merchandise which have earned that place did it by being dependable over years, not by delivery a intelligent AI characteristic. The merchandise that haven’t earned it but are at all times one good Claude dialog away from being pointless.
Belief and the human layer. As AI commoditizes outputs, the individual or agency that vouches for the end result features worth. Ruby Receptionists has spent over twenty years constructing a status for answering telephones with real human heat and authorized market data. That isn’t a characteristic that may be replicated at zero value. The identical is true for the advisor who has been inside your methods for eight years and is aware of the place each integration lives and why each resolution was made. (Cough, cough, Affinity Consulting). Earned credibility and human judgment should not version-one options. They accumulate. The merchandise and individuals who have gathered them are offering one thing AI can not undercut.
Scale and operational consistency. Doing one thing as soon as with AI is simple. Doing it the identical approach ten thousand occasions, with an entire audit path, throughout a two-hundred-attorney agency, with out a human reviewing every occasion—that could be a product. The instruments that operationalize AI relatively than merely exposing it are offering one thing a general-purpose mannequin can not: governance, repeatability, and organizational belief. The distinction between “we use Claude for this” and “we’ve got a system for this” is the distinction between a functionality and an operation. Enterprise-grade platforms that wrap AI in actual course of infrastructure are constructing on that hole.
What This Means for Your Shopping for Choices
The Claude Check isn’t just for evaluating what you’re about to buy. Run your present stack via it.
Not as a purpose to cancel every thing—workflow place and deep integrations should still justify the price of a product even when its AI options would fail the take a look at. However as a lens for sincere conversations with your self and your distributors. What am I really getting right here? If the AI layer disappeared tomorrow, would I nonetheless want this? What’s load-bearing and what’s simply handy?
The query value sitting with: when you rebuilt your tech stack from scratch at present, which merchandise would you pay for with out hesitation? These are your foundations. Those you’d hesitate on—that’s the take a look at doing its job.
Subscribe
Get skilled insights and sensible suggestions delivered to your inbox each week.
Can Your Agency Go the Claude Check for Your Shoppers?
The identical take a look at applies to your apply. Which components of what you do would a shopper battle to duplicate with a great AI system and a day? Which components of your work are loops dressed up as experience—pattern-matching that AI is coming for first?
The corporations that may reply that truthfully are those positioned to make good know-how selections. And to construct one thing that lasts.