The email arrives in your Slack channel within minutes. Three hours later—guaranteed—an attorney has marked up your commercial contract, flagged the problem clauses, and explained what needs to change before you sign. Total cost: $500, flat.
It's an audacious pitch, the kind that makes traditional BigLaw partners wince. But Ryan Walker and his co-founders at General Legal aren't trying to win over Cravath partners. They're after the startups drowning in legal bills that arrive weeks after the work, often with no warning about the final tab.
Walker knows the legal AI space intimately. As CTO of Casetext, he helped build some of the earliest generative AI tools designed specifically for attorneys—technology sophisticated enough that Thomson Reuters paid $650 million for the company in 2023. Now he's back with something more radical: not selling AI tools to lawyers, but running an entire law firm on AI leverage.
Since launching a private preview last month, General Legal claims to have pulled in over 20 growth-stage companies. All communicate through dedicated Slack channels. Most contract reviews, the firm says, come back in about an hour, though they guarantee three. The first review is free if you sign up before March 24, when General Legal presents at Y Combinator's Demo Day alongside two other AI-native law firms.
Whether this model can scale—and whether quality holds when the firm moves beyond its first few dozen clients—remains the central question.
Speed as Strategy
The economics are jarring if you're used to traditional legal billing. Contract reviews at established firms can easily run into the thousands, billed by the hour with timelines stretching across days or weeks. General Legal charges $500 for any contract between three and 50 pages, covering not just the initial review but every negotiation turn through final signature.
Shorter contracts cost $250. Anything over 50 pages runs $10 per additional page. Drafting from scratch: $2,000.
The speed comes from what the firm calls an "AI-native" workflow, though details remain somewhat opaque. A proprietary system named Sentinel performs the first pass—redlines, annotations, issue spotting—in roughly 10 seconds. Then licensed attorneys step in, reviewing Sentinel's work and applying what the firm describes as "strategic judgment and market context." According to a company blog post, attorneys typically spend 15 to 30 minutes completing their portion.
That's the firm's claimed "100× leverage" over traditional legal work, where junior associates might spend hours marking up standard commercial agreements, billable at rates north of $400 per hour.
Specialized bots monitor those Slack channels to trigger internal workflows. Attorneys use a custom Microsoft Word add-in during review. The infrastructure suggests significant engineering investment, though General Legal hasn't disclosed funding amounts beyond listing its investors: Y Combinator, Susa Ventures, AME Cloud Ventures, and BoxGroup.
Operating hours run 5 a.m. to 6 p.m. Pacific, Monday through Friday. Outside those windows, you're waiting. Video, phone, and email are available, but Slack is where most communication happens—instant, informal, and trackable.
The Casetext Pedigree
Walker isn't doing this alone. His co-founder Javed Qadrud-Din is a Harvard Law graduate who practiced at Fenwick & West—the go-to Silicon Valley firm—before leading AI development at Casetext and later working at Meta. The third co-founder, J.P. Mohler, also went to Harvard Law and put in time at Cooley and WilmerHale, both major players in startup legal work.
These aren't software engineers pretending to understand legal services. They're lawyers who got deep into AI and decided the practice model itself needed rebuilding.
Still, credentials only take you so far. The question is whether human attorneys, even with AI assistance, can maintain quality while moving this fast. And whether clients—particularly sophisticated startups accustomed to BigLaw polish—will trust the output.
General Legal has staked out a clear position here. In a February blog post titled "Claude Work is an Awesome Tool... but it Cannot Replace Lawyers," the firm tested Anthropic's Claude Work on a real contract and concluded that AI helps but human judgment remains essential. They even open-sourced a tool called "Find The Fuck Up" (yes, really), an LLM-powered contract error checker built on Claude Sonnet 4 that flags drafting mistakes with severity ratings.
The tool serves dual purposes: useful for clients, and marketing evidence that traditional legal work often contains the kinds of errors AI can catch.
A Crowded Field Takes Shape
General Legal isn't alone in trying to reinvent legal services around AI. Y Combinator's Winter 2026 batch includes three AI-native law firms, each carving out different verticals. Arcline targets startup legal work with same-day delivery goals—$300 for NDAs, $1,500 for MSAs. LegalOS specializes in immigration law, promising 48-hour filings and claiming a perfect approval rate so far.
The model is spreading. Covenant raised $4 million in seed funding last year to serve institutional investors with LP documents. Lawhive, operating in the UK and Arizona, just closed a $60 million Series B in February and employs roughly 500 lawyers on its platform—a scale General Legal is nowhere near yet. Story LLP and Shepherd offer subscription-based legal services. Eudia Counsel operates under Arizona's Alternative Business Structure program with $105 million in funding.
These firms differ fundamentally from legal AI software companies like Harvey, LegalOn, or Robin AI, which sell tools to existing law firms and in-house teams. The new wave positions itself as legal service providers that happen to use AI extensively, not software vendors hoping lawyers will adopt their platforms.
It's a distinction that matters. Regulatory scrutiny is increasing. The American Bar Association issued its first ethics guidance on generative AI last July, outlining duties around competence, confidentiality, and reasonable fees when lawyers use AI tools. The UK's Solicitors Regulation Authority approved Garfield AI last March as the first AI-driven law firm. Arizona's Alternative Business Structure program enables non-lawyer ownership of law firms that integrate AI—an arrangement still prohibited in most U.S. jurisdictions.
An AI Agent as Client
On February 14, General Legal claimed to be the first law firm hired by an AI agent. Using a service called Rent A Human, the firm accepted an inquiry from an AI agent and countersigned an engagement letter the agent had initiated.
It's a narrow claim—requiring that the AI agent initiate and "sign" on behalf of a human—but Mohler highlighted it on LinkedIn as an early example of machine-initiated B2B workflows. Questions about authorization, consent, and unauthorized practice of law when AI agents negotiate legal services remain unanswered. The founders seem interested in exploring those boundaries, though perhaps more as a proof of concept than a serious business line.
Still, it signals where they think procurement is heading.
The Scaling Question
General Legal's website currently shows a placeholder for "growth stage companies represented." No client logos, no testimonials—at least not yet. The claim of over 20 clients during private preview suggests initial traction, but it's early.
Details about the underlying technology stack remain limited. The firm's blog describes Sentinel as leveraging "state-of-the-art LLMs" without specifying whether these are proprietary models or orchestration of third-party APIs from OpenAI, Anthropic, or others. The custom Word add-in and Slack bot infrastructure clearly required engineering investment, but how much, and how sustainable the economics are at scale, remains unclear.
Pricing appears limited to per-contract fees at launch. Whether the firm will offer volume bundles or retainer arrangements for companies with heavy contracting needs hasn't been announced. And while $500 per contract sounds attractive, the math changes if you're negotiating dozens of deals per month.
The bet, ultimately, is that startups will pay fixed fees for speed and transparency over the traditional model of hourly billing with unpredictable timelines and surprise invoices. For growth-stage companies tired of legal bills that show up weeks after the work—and pricing conversations that require negotiation before the lawyer will even glance at a contract—$500 and three hours sounds refreshingly straightforward.
Whether that simplicity survives contact with scale is another matter entirely. The Casetext pedigree gives Walker and his co-founders credibility on both the AI and legal fronts. But scaling a law firm on AI leverage while maintaining attorney oversight and quality control will test whether the model can sustain margins at volume.
Perhaps the most telling detail: General Legal operates only during Pacific business hours, Monday through Friday. That's a constraint traditional firms don't have, particularly those with global teams. It's also a hint that, for now at least, the firm is managing capacity carefully—building the foundation before promising more than it can deliver.
Time, as always in startups, will tell whether General Legal's speed advantage holds as the client roster grows. For the moment, 20 companies seem willing to find out.
