Hello to all ~10,000 of you from around the globe. Thank you for reading and sharing!

Back to long form today.

TL;DR 2025 was one of Big Law's best years on record. That's precisely the problem. Older disruption narratives never landed in the partnership core because they were about the edges (e-discovery, managed services, process work). What's changing now is different. AI is exposing that the modern law firm was never one business. It is four: origination, governance, production, and accountability, each with a different economic profile and each under pressure from a different direction.

The firms that matter in 2030 will know which layer they are defending, which they are redesigning, and which they are quietly at risk of losing.

This is the map. 👇

Law's Genome

Why Big Law's next disruption is structural, not just technical

The most misleading idea in legal strategy right now is that Big Law has a disruption problem. It doesn't, at least not in the familiar sense. Elite firms have not been irrational to ignore most of the last decade's disruption talk. In 2025, the average U.S. law firm posted 13.0% profit growth; demand had one of its strongest years since the Global Financial Crisis; and worked rates, the rates firms actually charge and collect, not list rates, rose 7.3%. At the top of the market, rate power remained extraordinary: Reuters reported disclosed 2026 rates as high as $4,000 an hour for top partners at Susman Godfrey, with other elite firms not far behind.

That is exactly why older disruption narratives never landed in the partnership core. The old story was mostly about the edges: e-discovery, staffing, managed services, process-heavy work, and ALSPs. That market is real and growing. Reuters reported it reached $28.5 billion in 2025. But it rarely threatened the highest-priced, highest-status center of elite-firm economics. From the perspective of a premium firm, that first wave was important without being existential.

The more important change now is different. AI is not simply making legal work faster. It is exposing that the modern law firm is not one business. It is a bundled structure of distinct businesses that have long traveled together under one brand: the business of winning work, the business of governing risk, the business of producing work, and the business of standing behind the answer. Those layers still sit inside one partnership shell. But they no longer move together economically. That is the strategic issue.

Norm AI as diagnostic instrument

Norm AI matters right now not because it is the only serious legal-AI company, and not because it will somehow "replace" elite law firms. It matters because it is the clearest current x-ray of what a law firm actually is. Since November 2025, Norm has taken a new $50 million investment from Blackstone, launched the affiliated but independent New York firm Norm Law, added former Sidley executive committee chair Mike Schmidtberger and other senior Big Law hires, and launched a Legal AGI Lab focused on building "legal infrastructure for AI agents." Norm Law's own website now describes itself as 100% AI-native and says it has 60+ attorneys and legal engineers. One of the market's most watched legal-AI companies is no longer simply selling software into the existing firm model; it is beginning to redraw the boundary between software, legal engineering, and licensed practice.

Earlier bifurcations like Clearspire's two-company structure, Atrium's startup-backed hybrid that ended with layoffs in 2020, were mostly arguments about delivery structure. Norm is different. Its official materials describe a system for turning regulations, statutes, legal standards, and workflows into AI agents. That is not a better wrapper around a law firm. It is an attempt to extract and operationalize the logic inside legal work itself. When a company explicitly separates legal expertise into machine-executable systems, pairs those systems with a separately structured law firm, and frames the whole project as infrastructure for AI agents, it becomes much harder to keep pretending that the modern law firm is one integrated business.

The Genome

The firm's genome consists of its bundled logic of winning work, governing risk, producing work, and standing behind the answer. This is being read for the first time. And once read, what emerges is a stack we can categorize as origination, governance, production, and accountability.

Origination is not marketing in the narrow sense. It is the revenue-control layer of market sensing, relationship intelligence, client-specific opportunity recognition, cross-selling, positioning, pricing, and the ability to surface advisory value that firms routinely bury inside billable legal work. This layer has historically been under-owned because partners were expected to build books of business inside institutions that gave them mediocre business infrastructure.

That is changing. Intapp's April 2026 expansion of its Activator partnership with DCM Insights is one of the clearest market signals that partner growth itself is becoming a software category. Intapp says its Celeste agents can surface relationship signals, identify client engagement opportunities, and deliver prioritized next-best actions directly to partners; it points to research suggesting Activator behaviors can lift average partner revenue generation by up to 32%. Whatever one thinks of the branding, the implication is clear: the pre-matter layer is being productized.

The next legal-AI control point may not be the brief. It may be the brief before the brief. If production becomes cheaper and more systematized, the scarce asset shifts upstream toward noticing demand, framing and generating demand, and converting expertise into visible, purchasable value.

Capital has noticed. McDermott's reported exploration of an MSO structure, and the legislative reaction now underway in Illinois and California, suggest that investors and regulators alike understand that the old partnership bundle can be separated into distinct layers even if the formal practice of law cannot be directly owned. The traditional private equity tool of choice, the MSO, attacks the back office, which may work for consumer or volume-driven practices. Big Law's next opening is likely farther upstream: a new front-end operating layer for partner-level revenue generation. A future piece in this series will go deeper here.

Governance used to live in policies, committee memos, training decks, and human habit. AI is pushing it into infrastructure. Intapp's February 2026 partnership with Harvey is significant not because another vendor integration happened, but because Intapp said existing ethical-wall policies will sync automatically with Harvey's access and sharing controls across Assistant, Vault, and Workflows. Governance is no longer just what the firm says lawyers must do; it is increasingly what the system will and will not let them do. Reuters' reporting on proposed Federal Rule of Evidence 707, which would subject certain machine-generated evidence to reliability scrutiny, makes the shift harder to dismiss as vendor theater.

Production is the part of the story everyone sees first: research, drafting, due diligence, contract review, workflow automation. But the important shift is not that lawyers use AI. It is that production is becoming a system rather than a loose collection of lawyer tasks. Freshfields' new partnership with Anthropic is a good example. On April 23, Freshfields said it is deploying Claude across a platform serving 5,700 users, will receive early access to future models and tools, and will co-build legal-focused applications and agentic workflows for research, drafting, contract review, due diligence, and internal business-services processes. Reuters separately reported the partnership as part of a broader shift from experimentation toward scaled adoption and co-development. That is not a pilot. It is production architecture.

(The hidden machine)

Once production becomes systematized, a deeper collision surfaces; one that firms have sensed for three years without naming.

The old law-firm pyramid did three jobs at once. Junior work generated billable hours. Those same hours trained future lawyers. And that leverage produced partner margin. AI begins to pull those jobs apart. Clients buying output are less willing to subsidize apprenticeship. Firms can no longer assume training will pay for itself through leveraged human production. Outside capital is beginning to ask whether the non-lawyer layers around the practice can be separated and scaled on their own.

What looks like a talent problem is really a structural one. The profession is losing the hidden machine that once financed its own formation.

That leads to the fourth layer. Accountability is the layer automation does not remove. Production can be delegated; accountability cannot. Courts, clients, and regulators still look to lawyers and firms when something goes wrong. If anything, the more production is systematized, the more valuable this layer becomes.

Earlier this week, Sullivan & Cromwell apologized to a federal bankruptcy judge after a filing contained AI-generated inaccuracies, and the firm acknowledged that its own AI-use policies and secondary review procedures had not been followed. The incident is a reminder that responsibility still attaches to counsel, not to the model. The premium proposition of elite firms may therefore migrate away from raw first-draft throughput and toward judgment under ambiguity, exception handling, supervision, and the willingness to stand behind an answer in court, in a boardroom, or before a regulator.

One firm, four economies

Seen this way, the law firm of the near future is not a single business "using AI." It is a stack of functions with different automation profiles, data needs, and pricing logic.

The old partnership remains one institution from the outside, but its internal layers are no longer economically identical.

This is why the market's loudest debate, whether AI will replace lawyers, is too crude to be useful. The better question is which layer inside the firm is changing first, and who will own it.

Thomson Reuters' 2026 AI in Professional Services Report says two-thirds of corporate respondents want their outside firms to use AI, yet fewer than 20% mandate it, leaving firms and clients largely in the dark about one another's AI goals.

The ACC/Everlaw 2025 survey found that 64% of in-house counsel expect generative AI to reduce reliance on outside counsel and 50% expect lower outside-counsel costs. Clients are not uniformly prescribing the future, but they are clearly preparing to buy legal work more selectively.

Controlled separation, not collapse

The likeliest outcome is not collapse. It is controlled separation. Some functions will remain partnership-native because they depend on trust, privilege, judgment, and reputation. Some will become productized. Some will sit in affiliate or managed-service layers. Some will be bought from platforms. And some may migrate into new entity structures entirely. Reuters' reporting over the last year makes the direction hard to ignore: Arizona approved KPMG Law US, making KPMG the first Big Four firm able to practice law in the United States; Eudia Counsel launched as an AI-augmented Arizona law firm; and Norm AI launched an affiliated but independently structured New York law firm, Norm Law, alongside a new $50 million Blackstone investment. These are not the same model, but together they show the legal market experimenting with different separations between licensed practice, operating layer, and technology layer.

The threat to elite firms is not that premium practices suddenly disappear. It is that firms modernize the production layer while leaving the revenue-control layer underbuilt, the governance layer under-embedded, and the accountability layer under-defended. That would allow a premium institution to remain highly profitable while becoming strategically less integrated.

For firm leaders, the practical implication is not that they need to dismantle the traditional stack tomorrow. It is that they need to identify which parts of it still belong together. Two questions are worth answering before any AI roadmap:

The modern firm was never one homogeneous enterprise, or "platform," as many leaders like to call it. It was a successful bundle of different businesses traveling together under one brand. That bundle can remain powerful. But it will no longer remain strategically coherent by default.

Big Law is not being dismantled. It is being decoded. The firms that will matter in 2030 are already making these decisions in 2026, mostly without naming them. This newsletter exists to name them.

 Talk soon again, Josh.

To read previous editions, click here.

DISCLAIMER: None of this is legal advice. This newsletter is strictly educational and is not legal advice or a solicitation to buy or sell any assets or to make any legal decisions. Please /be careful and do your own research.