The scrambled competitive map

The Paradox

We had two reports open on screen, and they were telling us opposite things.

The first was the Thomson Reuters/Georgetown State of the Legal Market report. Record profits. Am Law 100 profits per lawyer up nearly 54% since 2019. Ninety percent of legal dollars still flowing through hourly billing. The traditional model isn't just surviving — it's thriving, apparently.

The second was the Harbor 2025 Law Department Survey. Sixty-five percent of corporate legal departments have intentionally retained work in-house over the past two years. Expectations of increased outside counsel spending have dropped from 58% to 37%. And eighty-five percent now have dedicated AI resources or committees — not exploring, deploying.

Both reports are credible. Both have solid methodology. And they point in opposite directions.

Right now, at least fifteen Am Law 100 firms are paying first-year law students up to $50,000 to spend their 1L summer doing public interest work — not at a competing firm — in exchange for a commitment to return as a second-summer associate. They are recruiting students before they've taken their first exams. How does that fit in a world where EY and the big consulting firms are cutting their graduate intake by double digits, and recent college graduates face the highest unemployment in a decade?

And the firms at the very top are not standing still either. Kirkland & Ellis — the highest-grossing firm in the world — is reported to be spending around $500 million to build artificial intelligence it will own outright: more than $100 million this year, hundreds of millions over three to four years, 180 technologists trained on how 250 of its own lawyers work, funded out of the firm's own revenue rather than licensed from anyone else. The most profitable incumbent in the market is raising rates on its clients and, at the same time, betting half a billion dollars of its own money that the way it works today will not hold.

Record profits, record recruitment spending, half-billion-dollar internal bets, and a simultaneous narrative of professional destruction. Most commentary resolves this by picking a side — the profession is fine, or the profession is doomed. We think the problem is the map. The map most commentators draw has two kinds of supplier — incumbents, traditional law firms who dominate, and "alternative" providers who challenge from the margins — and one kind of customer, under resourced corporate legal departments who, notwithstanding the success of the in-house counsel movement, will always be dependent on law firms for specialized expertise and alternative providers for efficiency. The real market has many more kinds of player than that and the lines between them are rapidly dissolving as the various actors keep changing which kind of player they are.

The Map Before AI

The pre-AI map had four positions and clear boundaries.

The incumbents were established law firms, primarily the Am Law 200 — hierarchical, partnership-based, premium-priced, structurally central to corporate life. A Sullivan & Cromwell opinion letter carried institutional weight no alternative provider could replicate. Rule 5.4 of the ABA Model Rules blocks anyone other than lawyers from owning or controlling law firms, with corresponding state laws prohibiting the unauthorized practice of law by anyone other than licensed attorney. Partnership capital structures limited what could be invested in delivery. The gates were guarded.

The challengers consisted of a collection of disparate providers, most of whom — e.g., legal process outsourcers, electronic discovery providers, flexible staffing companies like Axiom — targeting commodity or routine work, or specialized consulting firms offering legally adjacent services in areas such as deal valuation, litigation management, and investigations. The exception were the Big Four accountancy networks — PwC, Deloitte, EY, KPMG — which since the late 1990s (when, with Arthur Andersen, they were the Big Five) have attempted to compete directly with large law firms, and after surviving the accounting scandals and regulatory crackdown in the early years of the twenty first century that was supposed to have killed this ambition, were by the 2010s quietly rebuilding their legal networks. By 2012, PwC was promoting legal services in well over 100 countries. David's research at the Center on the Legal Profession found that law firms, by targeting only "bet the company" work, had "unwittingly ceded the battle for industry expertise to the Big Four." But unlike the 1990s, this time the Big Four weren't trying to be law firms; they were building something different — integrated solutions combining legal, consulting, tax, technology, and project management. Ironically, as David observed at the time: "By increasingly making themselves look like the Big Four in how they define and deliver legal services, large law firms may actually be hastening a world in which these global giants will no longer have to disguise their ambitions from regulators and the public." By the beginning of the millennium's second decade, this prediction had become reality as the Big Four have come out of the shadows to compete directly with established law firms around the world, including in the U.S.

The Big 4's integrated solutions model in turn played in to the evolving needs of in-house legal departments to produce "more for less." Beginning with the "inside counsel revolution" pioneered by Ben Heineman at GE in the 1990s, large companies were increasingly growing their internal legal departments in an attempt to reduce costs by buying legal services wholesale rather than retail. By the 2010s, many of GCs were turning to alternative providers as a way to further reduce costs. As the market emerged from the covid pandemic, the most sophisticated of these new GCs were using a combination of these strategies to both build internal expertise and utilize a variety of resources across the evolving "ecosystem" of alternative providers to drive down cost and gain access to integrated and multidisciplinary service offerings. As Tracey Yurko at Bridgewater Associates reported, she was constructing what she called a "virtual law firm" — segmenting work by complexity, partnering with PwC, cutting costs by more than 50%. Her assessment was blunt: "On balance, I have to give the advantage to the Big Four if (and for now that is an if) regulations change and allow them to practice law." Why? Because, she explained, "they can deliver a more integrated type of service."

Notwithstanding these efforts, however, most companies continued to devote 60% of their legal spend to external providers, with the overwhelming majority going to established law firms — a division of labor that, as David's research indicates, has remained remarkably consistent over time. Alternative providers were largely confined to commodity work, with even the Big Four making limited inroads beyond process-driven work like post-merger integration or regulatory compliance, particularly in the U.S. The covid crisis only accentuated this advantage, as clients turned to established firms to navigate the novel legal issues the crisis presented.

And while the regulatory regime that traditionally insulated lawyers from external competition has cracked, most notably in the U.K where for more than a decade virtually all restrictions on multidisciplinary partnerships and external investment in law firms have been eliminated, this regulatory opening has so far had relatively little impact on the competitive position of top law firms. Thus, while the Big Four are now all licensed to practice law in the U.K., and KPMG has now taken advantage of Arizona's Alternative Business Structure program, which eliminated Model Rule 5.4 and other restrictions, to become the first of the Big Four to legally practice law in the U.S., the year over year record profits recorded by top law firms demonstrates that the moat that Big Law has meticulously built to secure its premier competitive position in the corporate legal services market remains intact. The map was under stress. But the lines held — Rule 5.4, partnership capital, cultural resistance, and the absence of technology to deliver integrated solutions at scale. The four positions stayed in their boxes because the boundaries between them held. Then the boundaries began to dissolve.

The Players AI Added

The Silicon Valley frame for AI in legal has been a supply-side binary: copilot versus autopilot. Julien Bek, a partner at Sequoia Capital, put it as cleanly as anyone: "A copilot sells the tool. An autopilot sells the work." It is a useful cut, but it names only two of the new players — which is part of why the map has been so hard to hold still. Over the last two years or so several kinds of player have arrived.

The tool vendors sell software into existing firms. Harvey is the canonical example, alongside Legora, Thomson Reuters' CoCounsel, LexisNexis, and Spellbook. A lawyer's hours get more leverage, but the lawyer — and the firm, and the bill — stay in place. This is Bek's "sells the tool" pole, and it is the least disruptive kind of player on the map: the client relationship and the fee structure are left exactly where they were. Combined private valuations of the two leaders alone, Harvey and Legora, exceeded $16 billion by early 2026, with most of the Am Law 100 among their customers. They enhance the lawyer and preserve the firm.

The AI-native firms rebuild delivery from scratch. Covenant, a fund-formation firm with six lawyers, delivers limited-partnership-agreement reviews at roughly 90% below traditional pricing. Crosby is Sequoia- and Cooley-backed. Norm Ai raised more than $140 million from Bain Capital, Blackstone, Vanguard, Citi, and Marc Benioff, and recruited Mike Schmidtberger — after seven years chairing Sidley Austin's executive committee — to chair its new AI-native law firm, Norm Law; he told Bloomberg Law he "confronted the future." Manifest OS raised a $60 million Series A at a $750 million valuation and builds and runs its own AI-native firms rather than selling software to incumbents; its CEO, Dan Mishin, said it "made the hard choice to not sell our AI software to existing law firms who are often beholden to billing customers more hours as a means to better compensation." This is Bek's "sells the work" pole. What defines the type is the cost curve — technology costs scale logarithmically while human labour scales linearly — so a firm built to run on the former competes on a slope the incumbents cannot match.

The platform owners sit underneath both. On May 12, Anthropic launched Claude for Legal: a dozen practice-area plugins from M&A diligence to deposition prep, more than twenty connectors into the software firms already run on, and a Microsoft-365 agent that carries context across Word, Outlook, Excel, and PowerPoint. Live users on launch day included Freshfields (5,700 users across 33 offices), Quinn Emanuel, Holland & Knight, and Crosby Legal. Read the connector list carefully: Harvey, CoCounsel, and LexisNexis appear on it as integrations into Claude, not as platforms above it. When Anthropic's first legal plugins shipped earlier in the year, the market wiped out roughly $285 billion in software and professional-services value. This player is neither copilot nor autopilot; it is the substrate the others run on, which is why the tool vendors that once treated the model as raw material now appear as features inside it.

Anthropic was first into this tier but it has not stayed there alone, and the speed of the others arriving is its own kind of evidence. On April 30 Microsoft shipped a Legal Agent directly inside Word — not a general copilot pointed at a contract but a purpose-built agent for playbook-driven clause review and tracked-change redlining, built by the engineering team it had absorbed in January from Robin AI, a venture-backed legal-tech vendor that collapsed in late 2025 after failing to close its funding round. A tool vendor did not become a plugin on the platform; it died, and its engineers became the platform's legal feature.

A month later, on June 1, OpenAI stood up a dedicated legal vertical and hired Jason Boehmig — a former Fenwick lawyer who co-founded the contract-management company Ironclad and built it to a $3.2 billion valuation — to lead it. Within roughly six weeks all three of the largest model providers had planted a flag in legal, which tells you the platform layer is not a settled position one company holds but a contested one they are all now fighting over. And the contest folds back on itself: Claude for Legal reaches lawyers through a Microsoft-365 agent even as Microsoft ships a legal agent of its own, so the substrate and the application now compete inside the same Word document.

The adaptive incumbents are the category the standard disruption story leaves out. Clayton Christensen's framework saw only incumbents and challengers; it paid far less attention to the incumbents that adopts the challenger's attributes and survives. Freshfields going firm-wide on Claude is one version. Kirkland's half-billion-dollar build is the more aggressive one: rather than license what everyone else licenses, it is building and owning its own infrastructure, on the theory that at the top of the market the technology is too central to rent. In this respect it is trying to emulate IBM, which unlike minicomputer rivals like Wang and Digital Equipment Corporation, saw the personal computer revolution coming and made a commitment to disrupt itself by establishing its own PC division. Only unlike IBM, Kirkland is not hiding its disruptive innovation from powerful incumbents within its organization as IBM did when it located its PC division in a secret location in Boca Raton, Florida, away from the mainframe division which the CEO rightly feared would kill it. Instead, Kirkland is publicly embracing disruption as a way of enhancing its status as the world's leading incumbent law firm. How many firms are willing — or able — to follow Kirkland's example remains to be seen. Kirkland's extraordinary size and revenue — $10 billion in revenue, $11.1 million profit per partner — give it the ability to invest and leverage the returns over its existing platform that few other existing firms can match. But even firms that cannot match Freshfields' or Kirkland's or A&O Shearman's (which was Harvey's first major law-firm partner) investment, all firms are going to have to adapt to the new competitive map even within the Big Law legal market that these firms are creating. Not even traditional law firms are a static category. And treating it as if it is is how you miss how the map as a whole is scrambling.

There is a fifth kind of player taking shape alongside these — the deployment firms that sell neither tools nor work but the act of putting AI to work inside someone else's organisation, stood up in 2026 through multi-billion-dollar joint ventures between the foundation-model companies, private capital, and the big consultancies. Because what they compete for is the integrated, upmarket territory the profession itself is trying to move into, we take them up in the final piece rather than here.

And as we go to publish this essay, Kirkland has just announced a joint venture with Palantir to do just this. While the stated goal is to drive its own tools through its own work flows, the FT article announcing this development notes at the bottom that A&O Shearman has created its own suite of tools that it intends to package and sell, presumably to clients and possibly to other law firms. The map keeps scrambling as incumbents, disrupters, and new players try to figure out the best way to exploit the tremendous potential of law, as we argued in our prior essay, as a gateway drug to value creation and extraction in an increasingly VUCA world.

Capital Becomes a Player

There is one kind of player the supply-side map leaves out entirely, and it turns out to be moving fastest: the capital itself.

For a while the role of capital was familiar — venture money funding the new entrants. What is unusual now is how indiscriminate it is. Sequoia funds the tool vendors (Harvey), the AI-native firms (Crosby), the platform owner (Anthropic), and the deployment JV — every type on the map at once. When the market cannot decide which model wins, it funds all of them, which means the capital has no stake in the boundaries holding.

Private equity is the larger force, and it is not only a funder. Blackstone is, at the same time, a major law-firm client, a lead investor in Norm Ai, and a co-founder of Anthropic's deployment JV. It is paying the incumbent, funding the challenger, and building the thing that would replace both. And PE's real leverage runs through its portfolio: a fund that integrates AI legal delivery across the thousands of companies it owns reshapes the demand for legal services across a whole slice of the economy, without ever touching the regulatory question of who may practise law.

Most recently, Carta — the cap-table software company that reached roughly $500 million in annual recurring revenue last year — acquired Avantia, an AI-native law firm regulated by the UK's Solicitors Regulation Authority and therefore allowed to take on external investment that served more than 200 asset managers and, on its own account, 30% of the world's largest funds across some $15 trillion in assets. Carta has now launched Carta Law and become a law firm. Its CEO, Henry Ward, put the logic plainly: "The largest PE firms in the world are paying top-tier law firms for high-volume, ultimately routine legal work, and they shouldn't have to." Avantia's founder, James Sutton, had been making the same bet for years: "We founded Avantia in 2019 on a contrarian bet — that AI could deliver legal and compliance outcomes, not just assist with them."

This is the inversion the venture world has taken to calling "service as software" — software as a service turned inside out, where the service leads and software powers it — and it is the most extreme version of it so far. The threat is sharper than a single acquisition suggests. The firm Carta bought already served private-equity clients, arguably Big Law's most coveted clients. Owning it lets Carta insert itself directly into those clients' workflows, with an obvious path to move up from routine work into the higher-value advice that has always been the incumbents' fortress. The funder is no longer only funding new disrupters and providing premium fees for incumbent firms, it has started being a player itself.

And an even more direct market entry may be on the way. In May 2025, former A&O Senior Partners Wim Dejonghe and David Morley announced the formation of a new consultancy whose sole focus is to advise law firms about taking outside investment from PE Funds and other sources. In the intervening year, there has been rampant speculation about law firms that have expressed interest in taking PE investment. As the Carta example underscores, such investment is legal in the UK as well as in other important markets like Singapore. And while legal barriers remain in the U.S., litigation funders like Burford Capital, which already deploy hundreds of millions in investor capital to underwrite hundreds of matters a year, including in top law firms around the world, are pitching a "managed service" model in which law firms would create a separate PE backed entity to handle all of a firm's back office activity freeing up capital to deploy in its core legal business that would technically circumvent these restrictions.

In April, David hosted a Chatham House Rule Senior Leader Workshop on "Law as a New Investment Class?" in which 50 top managing partners, GCs, and potential outside investors came to discuss the desirability and feasibility of PE investment in law firms. And while there was both caution and skepticism — particularly from potential investors who expressed concern about what AI would do to Big Law's business model over the next few years — there was widespread consensus that the demand for capital to invest in new AI systems and to compete for talent will lead some law firms to welcome some form of PE investment. Kirkland's decision to invest $500 million in AI solutions will only fuel this interest among competitors who are unable to make this kind of investment on their own.

None of this is unique to law. Corgi, an AI-native insurance carrier — licensed in its own right, running underwriting and claims in-house rather than renting them from third parties — went from a Series A to a $1.3 billion valuation in about four months, and to roughly $2.6 billion a few weeks after that. And PE firms have been busily rolling up insurers, medical practices and hospital chains, and, most recently, accounting firms — the private-capital rollup of Grant Thornton, a firm just outside the Big Four, began in 2024 and gathered in another member firm this April. The pattern holds wherever a regulated service can be rebuilt AI-first: capital stops buying tools for the incumbents and starts assembling the provider itself. And law is the least concentrated nearly trillion dollar business in the world, with no single law firm holding much more than 1% of the market.

The Scramble: Categories Blurring

What scrambles the map is not the number of players. It is that the lines between them are dissolving, and the same actor will sit on several sides at once.

Each kind of player is eating into the one before it. The tool vendors, with $16 billion in combined private value, now appear as plugins on the platform they once treated as raw material. The AI-native firms are often built on that same platform — Crosby Legal was a launch customer of Claude for Legal, not a competitor to it. Even the big consultancies that the old Big-Four-versus-law-firm story would have picked as the eventual winners have, in several cases, stopped competing with the foundation models and taken equity in them instead — selling their own methodology branded as someone else's product, with the model-maker owning the layer underneath.

Look at any single actor and the category dissolves in your hands. Harvey sells into Am Law incumbents, runs a Shared Spaces product that puts firms and their clients in one workspace — infrastructure for the very disintermediation its customers fear — and is now a plugin on a platform that competes with it. Supplier, disintermediation tool, and distribution channel, all at once. Cooley, an Am Law firm, invests in Crosby, one of its potential replacements. The categories are not holding, because the players have no incentive to hold them.

This blurring of organisational categories is the defining feature of the moment, and not just in law. As David has argued for years the traditional categories of organization and thought that we have traditionally used to understand and define the world — public/private, global/local, law/business — are increasingly less useful (to the extent that they ever really were) as it is increasingly clear that the most important issues exist at the intersection of these competing narratives, with important truths on both sides of traditional dichotomies and new perspectives captured by neither. In other words, precisely the kind of multidimensional problems that Anthea captured in her book The Six Faces of Globalization and which led her to create Dragonfly to deploy AI to help decision makers make sense of this multi-perspectival reality.

And all of this is now happening on a clock that keeps speeding up. The half-life of any incumbent-or-challenger position is shortening: a firm that commits hard to one configuration can find it obsolete before the rollout is finished, which makes adaptation genuinely punishing rather than merely difficult. The average lifespan of an S&P 500 company has dropped from around 60 years in the mid-20^th century to just 15 years today. As the market for legal services becomes increasingly integrated into broader markets for business, technology, strategy, and geopolitics — precisely what gives law its central position as a gateway drug — it will also be subject to the same forces that are speeding up and destabilizing these other forces as well. Early movers carry the most exposure. Committing early to a particular tool or partner looks decisive at the time. But it can lock a firm into a bet the next turn of the technology renders wrong — and the people inside who championed it are the least able to walk it back.

This does not mean that the leading incumbents will necessarily be replaced by disrupters the way Kodak and big steel were replaced by digital cameras and mini mills. One reason the legal map scrambles rather than resolving, for example, the way a software market would is that the people inside law firms have far more power than their counterparts in most industries to slow, stop, or redirect change — through regulation, partnership structures, and professional culture. That friction, however, doesn't stop the reconfiguration; it just keeps it from settling.

The Demand Side Isn't Waiting

Every map we've drawn so far is a supply-side map. But the demand side isn't waiting for the supply side to settle.

Eric Dodson Greenberg, general counsel at Cox Media Group, described the shift in a Bloomberg Law essay that has stayed with us: "Every contract, negotiation, and filing feeds a system that captures not just what was done, but how it was approached — the reasoning, the tradeoffs, the institutional logic. That knowledge compounds independently of any individual lawyer. And crucially, it belongs to — and resides with — the client, not the firm."

Read that last sentence again: the knowledge belongs to the client.

In-house departments aren't watching the supply-side debate from the stands. Sixty-five percent are actively insourcing. Sixty-four percent expect AI to reduce their reliance on outside counsel. Eighty-five percent have dedicated AI governance committees. In-house use of generative AI more than doubled in a single year, from 23% to 52%, and is now by some estimates as high as 87%. As Bjarne Tellmann, former general counsel at Pearson and Haleon and our co-teacher on this course, puts it: clients have spent fifty years trying to figure out how to get out from under the stranglehold; AI gives them the tools to do it.

What's new is that the demand side now has a parallel supply channel that didn't exist a quarter ago. The deployment firms we set aside earlier — the foundation-model joint ventures we take up in the final piece — are built to reach companies that historically couldn't afford Big Three consulting fees to put AI to work. Goldman's Marc Nachmann calls it "democratising access" to that kind of capability. The in-house flywheel keeps spinning, now inside a restructured enterprise services market.

But this latest development also underscores why the demand side no longer belongs exclusively to the legal department. For years, companies have been trying wrest control over the legal budget from lawyers and put it through the standard procurement process. Legal has largely, although certainly not completely, been able to resist these efforts by claiming that "legal services are different" and that any effort to centralize legal procurement risks sacrificing quality in ways that potentially create significant risk. After all, the standard argument goes, law is what the economist call a "credence good," where the quality of the finished product cannot easily be predicted ex ante or determined ex post — especially by "non-lawyers" — and buyers therefore must rely on the reputation, credentials, and past experience of the service provider, the evaluation of which requires the special expertise of the legal department.

AI, and the big data on which it is based and which it makes available, challenges these assumptions. If, as Greenberg indicates, every law firm engagement creates a wealth of data about both process and outcomes than why can't clients begin to mine this data to evaluate providers based on predictions about the quality of the outcomes they produce as opposed to the inputs they provide — and then use actual outcomes to further refine their predictions and set prices on the basis of the value they receive? And if foundation companies are driving application of their models across a company's entire platform than other parts of the organization will theoretically be able to gain access to the same data-enhanced knowledge and understanding as the GC. After all, contracts do not exist entirely — or even primarily — within the legal department. Which is why many first generation contract lifecycle management companies began aggressively marketing to sales, marketing, compliance, and other functions which had both real budgets and P&Ls and were far more open to improving efficiency through technology than legal. As law becomes more central to business, business will seek to gain greater control over law.

Legal, of course, will fight to defend its own turf. And the fact that very centrality of law which is leading business leaders to try to gain more control over it has also increased the power and status of general counsels, many of which have now moved into even broader strategic roles within their organization, means that legal departments are likely to continue to be important players in a company's AI transformation. More friction in the gears — friction that will continue to scramble the map.

The Paradox Resolved — Or At Least Explained

Now the Thomson Reuters and Harbor reports both make sense.

Law firms are raising rates on captive clients — the highest-stakes, bet-the-company work where the opinion letter matters, where reputational guarantees can't be replicated, where clients need the firm's name and its unparalleled access to critical information and relationships that still exists largely in the heads of key partners. Am Law 100 profits are up nearly 54% since 2019, with 2025 being the best year ever for many top firms. That's real.

But Am Law 100 firms couldn't crack 2% demand growth in late 2025, while midsize firms surged to nearly 5%. General counsel are shifting routine work — and increasingly even moderately complex work — downstream. They're insourcing, building AI infrastructure, and using alternative providers, who they no longer view as "alternatives" but as preferable substitutes for law firms for a growing array of work. The margin compression hasn't fully hit the top yet because the top is raising prices on the work that can't leave, largely by poaching top talent. Lateral partner hiring hit a five-year high in 2025, with a disproportionate share going to Am Law 50 firms. As Kirkland latest moves underscore, returns to scale are escalating as the top of the top continues to pull away from even the rest of the top. Wachtell and Kirkland profit per partner — $11 to $12 million — runs roughly 50% higher than #10 Milbank ($7.6M), about double #20 Weil ($5.5M), roughly four times #50 Foley ($2.9M), and more than fifteen times #100 Wilson Elser ($677K). Gross revenue tells a similar story, with Kirkland's $10 billion almost 3 times higher than Simpson's $3.5 billion, and dwarfing traditional powerhouse Cravath, whose roughly $1.3 billion ranks only around fiftieth by gross revenue.

Whether these profit margins are sustainable is an open question. As Christensen noted a quarter century ago, margins often peak just before the structural break. Client anger at billing rates quickly climbing past $4000 for partners — and even $1000 for associates — is palpable. But even if some law firms are able to maintain these astronomical fees, it is likely that the space for those at the very top of the pyramid is narrowing.

Record profits and a client base quietly building exit infrastructure are both true, at the same time, in the same market. The fortress is real. The bags being packed are also real. If you only look at the supply-side binary, you see the fortress. If you look at the whole map — incumbents and the adaptive incumbents pulling away from them, the Big Four, the new AI-era players, capital that has become a provider in its own right, a demand side building its own exits — you see the fortress and the packing.

What the Scrambled Map Means for This Series

The scrambled map doesn't tell the profession where to go. What it does is clarify the dynamics that make prediction so hard — and point to what the rest of this series needs to address.

We've now mapped the landscape from several angles — law's structural centrality, platform strategy, the experiential gap, and the competitive dynamics reshaping the field. In our next piece, we turn to the resource every player on this map depends on and none of them is producing: domain expertise, and the training crisis that threatens it.

After that, we'll explore where the profession goes when the base is automated — moving upmarket, going "up and out" toward integrated solutions, becoming orchestrators rather than producers. Law firms moving upmarket aren't heading for unoccupied territory. The Big Four are already there. So are the new foundation-model deployment ventures, built on an AI cost structure that carries none of the partnership overhead. General counsel are building their own orchestration capability. AI-native firms are improving from below.

The old directions — stay put or move up — may both be harder than they look.