I have a prediction that I want to share with you. This is something that I envision happening just a few short weeks from now. I imagine seeing an associate at a law firm doing something that will make every product manager at Thomson Reuters and LexisNexis choke on their morning coffee. She has a contract dispute question. A Real one. There will be a partner waiting. And the clock is ticking.
She won’t open Westlaw. She won’t open Lexis. She won’t open her browser at all.
She types her question into a work-approved AI chat window. Twenty-four minutes later she has a memo, citations included, sent off to the partner. Entered 0.4 hours on her time entry system. And she is done.
The KeyCite red flag that Thomson Reuters spent generations building? Never saw it. The Shepard’s signal? Didn’t see that either. The annotated treatise hierarchy that some editor in Eagan, Minnesota agonized over? Came through as a flat blob of text in a JSON response that the model summarized into a single sentence.
Don’t get me wrong. Westlaw and Lexis were in the research process. She just didn’t noticed they were there. And that, friends, is the near-future I want to talk about.
I’ve been privately calling this “Shadow UX.” Think of it as the user-experience cousin of Shadow IT. We all know the effects of Shadow IT, right? That was when the marketing team started using Dropbox without telling the IT department, and three years later IT realized the entire company’s roadmap was sitting on someone’s personal account. Shadow UX is the same thing at the interface layer. An unauthorized layer sitting between the user and the vendor’s product, and the vendor doesn’t control it, doesn’t design it, and increasingly doesn’t even know it exists.
For legal information vendors, the Shadow UX layer is mostly an LLM with a few tool calls bolted on. There are other versions out there too: browser extensions that re-skin search results, paralegals building Notion dashboards off APIs, scraping wrappers feeding firm intranets. The AI agent is the one eating everyone’s lunch though.
Here’s why I’ve been thinking about Shadow UX so much lately.
For thirty years vendors competed on the browser/dashboard. The fancy charts. The little visual icons. The hover states. Pixel-perfect interfaces designed for a human eye scanning a screen. In 2026, the user is increasingly something else. It’s a model reading a JSON schema at inference time. If you’ve optimized your product for an audience that’s becoming the minority of your traffic, you’re going to find out the hard way.
OK so why now? Three things had to happen at the same time, and they all did within about eighteen months.
First, the models actually got good. The 2024 models couldn’t handle jurisdictional nuance. The 2026 models draft memos that pass partner review. Not every time, sure, though often enough that associates are using them anyway.
Next, the billable hour math became impossible to ignore. We bill in six minute increments. Any tool that turns a ninety minute task into nine minutes is going to get used, with or without IT’s blessing. (Sound familiar? Hello again, Shadow IT.)
And Finally, the Model Context Protocol showed up. MCP is the part of this story that doesn’t get enough attention. Imagine if every database, every research platform, every internal wiki spoke a common language to AI agents. That’s MCP. Companies like NetDocuments and Midpage adopted it. Specialized vendors are rolling out MCP servers for everything from patent search to legislative tracking. Once the protocol got standardized, the vendor’s UI stopped being a moat and started being a speed bump.
Now here’s the part that should worry legal information providers. The editorial work that built these companies, the headnotes, the Key Number system, KeyCite, Shepard’s, all of that gets flattened.
In a portal, a KeyCite red flag is loud. It’s red. It’s literally a flag. You see it before you see anything else on the page. In the Shadow UX layer, it’s a token in a JSON field. If the model’s summarization logic doesn’t promote it, the user never sees it. The signal is technically still there. It’s just invisible.
The headnote tree is worse. Editors spent generations nesting these things to show legal relationships. Models hate hierarchies. They flatten them into bullet lists, or worse, into prose. The categorical context disappears.
And then there’s the provenance problem, which is the one that actually worries me. When an agent synthesizes ten cases into one paragraph, the user gets a confident narrative. They don’t see that eight came from KeyCite-validated sources and two came from a sketchy public database the model decided to trust. The vendor’s brand was always the proxy for “this is reliable.” When the brand is invisible, the proxy is gone.
I’ll put it bluntly. If you’re a research vendor, your brand value is currently being laundered through someone else’s chat interface, and you’re not getting credit for it.
The pricing model is the other shoe about to drop.
Seat-based pricing is the deal we’ve all lived with since the 90s. You pay per lawyer. The lawyer logs in. Everybody understands. Now… an AI agent doesn’t log in. It doesn’t have a seat. It can do the work of fifteen associates in an afternoon though. So vendors are watching seat counts flatten while their compute costs spike. The infrastructure bill goes up while the revenue line goes sideways. That’s not a sustainable shape.
The industry is wobbling toward usage-based and outcome-based pricing. Pay per query. Pay per resolved research task. Pay per drafted clause. Salesforce and Zendesk are already doing this in their own categories. The math makes sense for vendors. The problem is that law firms hate metered bills. CIOs cite cost forecasting as the number one headache with consumption pricing. Nobody wants their Westlaw bill to look like an AWS invoice.
Here’s where the real fight is going to happen, and I haven’t seen anybody talk about it openly yet.
Put yourself in the chair of a Westlaw or Lexis sales VP. You’re watching seat utilization drop. Associates are logging in less. Partners barely log in at all. The minutes-per-seat metric you’ve been using internally to justify renewals is collapsing. Meanwhile your compute costs are spiking because the firm’s MCP-connected agents are hammering your APIs and MCPs at three in the morning to draft research memos.
What do you do?
I’ll tell you what you do. You add an AI agent access fee on top of the seat license. Premium tier. “Enterprise agentic access.” Whatever the marketing team lands on. And you keep raising the per-seat price every renewal cycle. Because if each seat is getting cheaper for the firm to actually use, your only path to flat or growing revenue is to charge more for each one. Double dip. Seats plus agents. Stack them.
Now flip the chair. You’re a firm CIO or a law firm library director. Your usage data shows seat logins dropping. Your associates are no longer going directly to Westlaw or Lexis. The vendor calls to renew, the price per seat is up 10%, and now there’s a separate line item for “agent access” that wasn’t on last year’s quote. You ask why you’re paying more for less. The vendor explains, with a straight face, that the value sits in the data, the agent extracts more value per query, and the bill reflects that. You disagree.
That’s the battle.
Firms have leverage they haven’t quite figured out how to use yet. If a general-purpose model with a good MCP integration can produce a defensible memo using free public data plus a couple of specialized vendor pipes, the firm doesn’t need a full Westlaw or Lexis subscription anymore. They need a few targeted pipes. Maybe federal cases, maybe a particular state’s regulatory feed, maybe a specialized treatise. The bundled subscription that’s been the vendors’ moat for thirty years is exactly the thing the agentic ecosystem can unbundle.
The firm-side response is going to come in three flavors. First, the headcount-only firms: “we’ll keep paying for seats at last year’s rate, take it or leave it, and we’re not paying a separate agent fee on top.” Second, the audit-and-cut firms: “show us actual usage data, justify the renewal price against actual logins, or we cut the seat count to match.” Third, the route-around firms: “we’ll keep a small premium subscription for the editorial signals we can’t get anywhere else, and we’ll point our agents at public data plus a few targeted MCP feeds for everything else.” Each of those is a different kind of headache for the vendor, and each one has a different ceiling on what the vendor can actually charge.
The vendors who win this fight will be the ones who can credibly argue their MCP server delivers something the agent can’t get anywhere else. KeyCite citator data, validated public-records overlays, proprietary treatises, expert witness analytics, the stuff that took fifty years of editorial labor to assemble. That’s the moat that survives. The vendors who try to hold the line on seat prices while gating their best data behind a separate agent fee will find their customers routing around them, because at that point the firm just buys the agent-access tier and treats the seats as a courtesy login for partners who still like the old interface.
My prediction: the first major firm to publicly announce they’re cutting Westlaw or Lexis seat count by 40% while keeping their MCP-tier subscription will set off an industry panic. Somebody is going to do this. Watch for it.
Now let’s talk about the verification tax, because this is where the AI evangelists get quiet.
The better the models get, the harder they are to audit. Sounds backwards, although it’s true. When errors are common, you spot them. When errors are rare, you stop looking, and that’s exactly when one slips through and ends up in a brief. There’s actual statistics on this. Researchers proved that the cost of estimating calibration error grows as models improve. For a ten-step agent loop, the verification cost can be a thousand times the cost of a single-step model. Lovely.
Then there’s what MIT called the Confidence Paradox. Models use more confident language when they’re hallucinating than when they’re stating facts. Thirty-four percent more, according to their 2025 work. So the smoothest, most reassuring chunk of your AI memo? Statistically, that’s the part most likely to be wrong.
Friends, this is a malpractice waiting room.
The worst version of the Verification Tax happens when associates trust the agent because it sounded confident, partners trust the associate because the memo looks clean, and the bar trusts the firm because nothing got flagged. Mariana Trench of false confidence. Somebody is going to get sanctioned, and the case is going to read like a horror story.
So what should the vendors do? I get asked this a lot lately, and my answer probably annoys them.
Throw out the human-first (or at least human-only) design playbook. The audience is the model now.
That sounds heretical to a UX designer, I know. Every editorial signal needs to be a structured field in the response payload. KeyCite/Shepard status should be a typed enum with a confidence score and a direct citation to the underlying authority. The model can then promote that signal in the summary, because it’s data instead of decoration.
Ship real MCP servers. Press releases about “AI partnerships” don’t count. Actual production grade tool surfaces with rate limits, auth, and schemas the model can read at inference. If you’re not in the agent’s toolbox, you’re not in the workflow.
Build provenance into every response. Hash-pinned citations. Quote spans with character offsets. Per-claim source attribution. Make hallucination expensive for the model to generate and easy for the human to detect. This turns the Verification Tax from a tax on the user into a feature for the vendor.
And reprice. Just reprice. Seat pricing is over. The metering infrastructure you’ll have to build is annoying, although it’s the only way the math closes.
The Legalweek 2026 lineup told us where the incumbents have landed. Thomson Reuters rebuilt CoCounsel on the Claude Agent SDK and is trying to “own the shadow” by being the agent itself. LexisNexis is leaning the other way, embedding Cowork into Protégé and treating Lexis as the “primary connection point” for content. Two strategies, same underlying bet, which is that lawyers will choose a curated agentic environment over a general-purpose model with specialized pipes.
I’m not sure they will. Lawyers use what works. If the general purpose model with a good MCP integration produces a better memo in less time, the walled garden becomes a walled relic.
If you don’t create solid MCP integration, your users will create workarounds that will get them what they need. I haven’t even scratched the surface of things like Codex Computer Use or Perplexity Computer in this article. But, trust me, tools like that will make it very easy for creative lawyers to just have the AI interact with the legal information. It’s just too much to try to cover here, but at least I’ll mention it for those vendors who think they control all the access points to their product.
For the lawyers reading this, here’s my unsolicited advice. Audit the grounding. When the AI summarizes a case, ask explicitly whether it checked subsequent treatment, and demand it surface the KeyCite or Shepard’s signal verbatim. Verify the pipes. Know which sources your agent is actually calling, because a “research result” from a web-search plugin and a research result from an MCP-connected professional database are very different animals. And keep the judgment. The application of law to fact and the strategic counsel you give a client cannot be delegated. That’s the part of the workflow that has to stay outside the shadow.
By 2030 the dashboard is dead. In fact, the browser may be dead. The most important UX hire at a major legal information vendor won’t be drawing pixels. She’ll be writing tool descriptions and tuning system prompts so an agent representing a lawyer she’ll never meet can find, validate, and cite the right authority on the first call. The portal becomes the back office. The pipe becomes the product.
Some vendors will accept this and build the best possible pipes. Those vendors will keep the editorial moat and figure out how to charge for it in a usage-based world. The ones who refuse will end up as the “Intel Inside” of legal research. Real. Important. Invisible. Priced like a commodity.
So here’s where I land. Shadow UX isn’t coming. It’s already in your firm, right now, and it’s growing. The interface your customers actually use is one you didn’t build, and the experience you spent decades polishing is being rendered, badly, through somebody else’s chat window.
You can’t fight it. The remaining choice is whether you’d rather be a great pipe or an irrelevant portal.
What are you going to do about it?