Your Lawyer Finished in 3 Hours. Why Is the Bill Still for 20?

Harvey AI — a company almost nobody had heard of two years ago. By end of 2024, revenue was just $50 million. By end of 2025, it had surged to $190 million. Nearly quadrupled in a year. Valuation rocketed from $3 billion to $11 billion.

This isn't another story of a Silicon Valley unicorn burning cash to grab market share. Harvey AI's value proposition is clear: help lawyers with legal research, contract review, and due diligence. Over 100,000 lawyers currently use its system. Among the AmLaw 100 — the top 100 U.S. law firms — 42% are already clients.

On the other side, Thomson Reuters spent $650 million acquiring Casetext and rebuilt it as CoCounsel. By early 2025, over 20,000 law firms and corporate legal departments were using it.

The American Bar Association's survey paints an even clearer picture: AI adoption at large law firms was 0% in 2022. By 2025: 46%. Three years, from zero to nearly half. The lawyer you hired is very likely already using it. Question is: do you know?

---

Here's a logic completely different from EP1. Chegg died because free 80% killed paid 95%. The legal industry is different. You're not going to use ChatGPT to fight a lawsuit. Legal services have a moat Chegg never had: the trust barrier.

So AI won't "kill" lawyers. But it's doing something far more subtle. Lawyers who deeply use Harvey AI save roughly 37 hours per month. At BigLaw partner billing rates of $800-1,500 per hour, that's approximately $30,000-55,000 of work per month.

Here's the question: will lawyers pass those time savings on to clients?

The reality is, the legal industry's business model is built on billable hours. In this model, efficiency gains aren't a good thing — they're a threat. If you use AI to compress a 20-hour contract review into 3 hours, you have two choices: bill the client for 3 hours (more honest), or bill for 15 hours (more realistic).

To date, no major law firm has publicly admitted to reducing billing rates because of AI. But 59% of corporate legal departments have said they "expect" the law firms they hire to use AI.

---

The first to get hurt aren't the partners — it's junior lawyers. What Harvey AI can do is precisely what junior lawyers and paralegals traditionally did: legal research, document review, contract drafting, case organization.

Between 2024 and 2025, white-collar hiring declined 12.7% overall. But for junior positions alone, the decline was 31%. Baker McKenzie, citing AI restructuring, cut 600-1,000 support positions — roughly 10% of its global workforce.

PwC data shows lawyers with AI skills earn 49% more than their peers. This isn't evenly distributed efficiency gains — it's K-shaped divergence.

---

EP1's Chegg was at least killed outright — you can sympathize with it. EP2 has more victims, but nobody dies. The group making money learned how to use AI to make even more, while those already at a disadvantage lost their only upward pathway. This is perhaps the most precise metaphor for the AI era: technology makes those at the top more efficient, and those at the bottom more redundant. The middle isn't a transition zone — it's a cliff.

---

_(Data sources: Harvey AI public funding disclosures and Forbes/Sacra reporting, Thomson Reuters CoCounsel product documentation, American Bar Association 2025 AI Survey, PwC AI skills premium report, stock data from Bloomberg/Yahoo Finance. Corrections welcome if any data errors are found.)_

_—Kinney's Wonderland_