AI Case Studies  /  Case study

Inside the Rise of AI-Native Law Firms

A new class of law firm is building AI agents into the work first and hiring lawyers to supervise them. Here is what Norm Law, Crosby, Garfield, and Manifest actually do, and what their model signals for pricing, staffing, and partner value.

By Anthony Guerriero  ·  Reviewed by The Leveraged Years Editorial Desk  ·  Published July 2, 2026

AI-native law firms build AI agents into the legal work first and put licensed attorneys on top to review and sign off, rather than bolting AI onto a traditional practice. Norm Law, launched with a $50 million Blackstone investment in the Norm Ai platform, runs "legal engineering," where lawyers build the agents that draft the first version of institutional financial work. Crosby, a New York firm backed by $85.8 million in venture funding, reviews commercial contracts for clients like Cursor at fixed per-document prices. In the UK, Garfield AI became the first purely AI-based firm the Solicitors Regulation Authority authorized, and in May 2026 it won a county court trial in which its client paid roughly £400 to recover £7,000.

A former Big Law chair walked into an AI company and did not walk back out

In January 2026, Mike Schmidtberger did something that made a lot of managing partners look up from their desks. After 35 years at Sidley Austin, where he ran the New York office and chaired the executive committee from 2018 to 2025 while the firm's annual revenue climbed from $2 billion to $4 billion, he joined a firm that did not exist eighteen months earlier. That firm is Norm Law, and its whole premise is that AI agents should do the first pass of the legal work while lawyers supervise. In his own words, he "walked through the doors at Norm Ai and confronted the future," per Bloomberg Law.

By April, Norm Law was still hiring aggressively, adding an ex-Sidley partner to lead its real estate team, according to Law.com. The point of a case study is not the headline hire. It is the model underneath it, and that model is now showing up in several real firms at once. If you want to understand where legal work is going before it arrives at your practice, these are the firms worth studying. For the working attorney trying to get ahead of it, our [Leveraged Attorney course](/leveraged-attorney) breaks the same shift into skills you can use this quarter.

What "AI-native" actually means, stated plainly

The phrase gets thrown around loosely, so here is the honest definition based on how these firms describe themselves. A traditional firm that adopts AI keeps its old structure and hands lawyers better tools. We covered that pattern in [How Law Firms Run on AI](/how-law-firms-run-on-ai). An AI-native firm does the opposite. It builds the AI agent into the workflow first, then staffs licensed attorneys around it to review, approve, and take responsibility for what goes out the door.

Norm Law is the clearest example of the institutional version. It was launched after Blackstone invested $50 million into the Norm Ai platform to build what the company calls Legal AGI, and it pairs AI engineers, "legal engineers," and attorneys on the same matter. Legal engineering, in Norm Ai's telling, is a discipline where lawyers build the AI agents that power the service instead of only drafting documents by hand. The firm targets financial services clients and is being built in collaboration with Blackstone's own in-house legal team, per PR Newswire. That last detail matters. The buyer helped design the product.

The venture-backed contract shop: Crosby

Crosby is the model built for speed on a narrow, high-volume task. Founded in 2024 by Ryan Daniels and John Sarihan, the New York firm pairs licensed attorneys with proprietary AI to review commercial contracts at fixed per-document pricing. Clients include the AI code editor Cursor along with go-to-market companies Clay and Unify, who use Crosby to close deals faster, according to Forbes.

The funding tells you investors believe the volume is there. Crosby raised a $5.8 million seed led by Sequoia in June 2025 and a $20 million Series A co-led by Index Ventures and Bain Capital Ventures with global firm Cooley participating, per Artificial Lawyer. Its April 2026 Series B of $60 million, co-led by Lux Capital and Index Ventures, brought the total to $85.8 million, and over the same stretch Crosby says the negotiated contract volume running through it grew from $30 million to just over $1 billion, per the Global Legal Post. Multiple outlets reported the Series B at a valuation of roughly $400 million. A traditional firm cannot scale a contract review team that fast without hiring hundreds of associates. That is the whole argument.

The regulated pioneer: Garfield AI in the UK

Garfield AI is the case study for what happens when a regulator actually blesses this model. In May 2025 the Solicitors Regulation Authority authorized Garfield.Law Ltd, the first purely AI-based firm it had cleared to provide regulated legal services in England and Wales, per the SRA. The firm, co-founded by former City litigation partner Philip Young and physicist Daniel Long, offers small businesses an AI assistant that guides them through the small claims process to recover unpaid debts.

Then came the part that made the traditional bar pay attention. After a three-hour trial at Wandsworth County Court on 14 May 2026, a Garfield AI client won a debt claim, recovering £7,000 while paying roughly £400 in Garfield fees. The defendant, by contrast, instructed both a solicitor and a barrister, according to Computer Weekly. Garfield reports it has recovered more than £500,000 across 600-plus claims. Read the SRA conditions closely, though, because they define the honest limit of the model: Garfield is not autonomous and acts only where the client approves each step, it does not propose case law because that is where models hallucinate, and named human solicitors remain fully accountable for every output.

The billable-hour killer: Manifest OS

Manifest OS is the case study for pricing. In April 2026 it raised a $60 million Series A at a $750 million valuation, reported as the largest Series A in legal technology history, with Menlo Ventures, Kleiner Perkins, and First Round Capital participating, per Legal IT Insider. Manifest builds and operates its own AI-native firms under the Manifest Law brand, starting in business immigration under Arizona's alternative business structure framework, with outcomes-based fixed pricing designed to end the billable hour. The platform runs client communication, drafting, research, billing, and reporting with human-supervised agents embedded in the workflow.

Worth noting the platform layer that sits underneath a lot of this. Legora, the Swedish legal AI company, reached a $5.6 billion valuation in 2026 on more than $100 million in annual recurring revenue, with its tools used across more than 800 firms and in-house teams, per TechCrunch. Legora is not itself a law firm. It sells the software that firms, AI-native and traditional alike, run on. The distinction matters when you decide which side of the market you want to compete on.

What this signals for pricing, staffing, and partner value

Three honest takeaways for a working lawyer, none of them hype.

On pricing, the billable hour is under real pressure on predictable work. When an agent drafts the first version of a contract or a debt claim in minutes, a client stops believing that six hours of a mid-level associate is a fair unit of value. Crosby prices per document, Manifest prices per outcome, and Garfield prices a case at a flat fee that is a fraction of a defended matter. That does not erase the hour everywhere. Complex, adversarial, and bet-the-company work still resists it. But the routine middle of the market is moving to flat and subscription pricing, and that is where most firm revenue actually lives.

On staffing, the pyramid changes shape. If an agent does the first draft, a firm needs fewer junior lawyers doing volume review and more senior lawyers who can supervise, plus a new role, the legal engineer, who builds and tests the agents. Norm Ai says it employs more than 30 attorneys doing exactly that kind of building. The uncomfortable version of this is that the traditional path, where you learn by grinding through document review for years, is the part AI replaces first.

On partner value, the Schmidtberger move is the tell. Deep client relationships, judgment on hard calls, and accountability for the final answer become more valuable, not less, because the machine cannot own the risk. What loses value is being the person who is simply faster or more thorough at producing a first draft. If you want a structured way to build the skills that keep a lawyer valuable in this model, that is the entire premise of the [Leveraged Attorney course](/leveraged-attorney), and you can [find your fit in about two minutes](/quiz) if you are not sure where to start.

Browse all AI Case Studies

Informational analysis for working professionals, not professional advice.