Legal AI Timeline
How a few short years of legal AI led to the moment — within two weeks in May, 2026 — that reframed the conversation for every firm.
How we got here
A few short years of legal AI — and why, within two weeks in May, 2026, one firm’s $500M bet and Claude launching legal-specific tools changed the conversation.
- Nov 2022
ChatGPT arrives. Overnight, every managing partner starts asking what generative AI means for the firm.
- 2022
Harvey is founded — OpenAI backs it as a seed investor — and starts building legal AI in parallel from day one.
- Feb 2023
Allen & Overy rolls Harvey out to 3,500+ lawyers across 40+ offices — the first firmwide generative-AI deployment in BigLaw, and the industry’s wake-up call.
- Mar 2023
GPT-4 launches — a step-change in reasoning that makes serious legal work suddenly feasible.
- Mar 2023
Casetext launches CoCounsel, the first GPT-4 legal assistant — proof the technology can do real legal work.
- Jun 2023
Mata v. Avianca: a federal judge sanctions lawyers who filed a brief full of AI-hallucinated cases — the profession’s first hard lesson in AI’s risks.
- Jun 2023
Thomson Reuters acquires Casetext for $650M — Big Legal's first major AI bet.
- Late 2023
The incumbents respond: LexisNexis ships Lexis+ AI and Westlaw launches AI-Assisted Research, bringing generative AI to the tools firms already pay for.
- 2024
Legal AI shifts from “build” to “buy” — firms pilot Harvey, Legora, CoCounsel and the rest instead of building their own.
- 2025
The lesson sharpens: data becomes the moat, not the model — a firm’s proprietary work product, not the underlying LLM, is the edge.
- Aug 2025
Harvey crosses $100M in annual recurring revenue with 50 of the AmLaw 100 as clients — adoption is no longer experimental, it’s operational.
- Nov 2025
Clio acquires vLex and its Vincent AI for $1B — the largest deal in legal-tech history, folding AI research and drafting into the platform thousands of smaller firms use.
- Mar 2026
Harvey raises again at an $11B valuation — investor conviction that legal AI is a generational market, not a feature.
- May 12, 2026
Anthropic goes all-in on legal — launching Claude for Legal with 20+ connectors and 12 practice-area plugins, its biggest move yet against Harvey and the incumbents.
- May 28, 2026
BigLaw’s big AI day: Kirkland & Ellis commits $500M to build its own proprietary AI platform — the largest AI bet by a law firm to date — while Fried Frank launches FundAssist for its funds practice the same day.
- June 2026
OpenAI launches a dedicated legal vertical, hiring Ironclad co-founder Jason Boehmig to lead it.
Questions it raises
The strategic questions this moment forces onto every managing partner’s desk.
- 1
Do we buy, build, or orchestrate?
Kirkland is betting $500M on build. Harvey’s own pivot says orchestrate best-of-breed models. Most firms can’t out-engineer a frontier lab — so what’s actually right for us?
- 2
Why stay in a closed, proprietary environment if we can just use Claude?
When frontier tools ship legal-ready out of the box, the case for spending years and millions on a walled garden gets harder to make.
- 3
Is flat-rate, value-based pricing the future?
If AI drafts the first version in minutes, the billable hour starts to look like a tax on inefficiency. Do clients still pay for time — or for outcomes?
- 4
What’s our actual moat?
If a competitor can stand up Claude’s legal tools in a weekend, the differentiator isn’t the model — it’s our proprietary work product, judgment, and client relationships.
- 5
What happens to the leverage model?
If AI does the first-draft work, the associate pyramid — and the economics underneath it — start to shift. Do we need as many hands?
- 6
Can we afford to wait?
This much moved in two weeks. The real risk may not be moving too fast — it’s standing still while rivals reset client expectations.
- 7
What is the role of the junior lawyer?
For decades, associates learned through research, drafting, and document review. If AI performs much of that work, how do firms train the next generation of partners? The leverage question is ultimately a talent-development question.
- 8
Will clients choose firms based on AI capability?
Historically, clients selected firms based on reputation, expertise, and relationships. Five years from now, will AI infrastructure, proprietary knowledge systems, and turnaround speed become factors in beauty contests and RFPs?
- 9
Who captures the economics created by AI?
If a task that once required 20 hours now requires two, someone captures that value. Will it be clients through lower fees, firms through higher margins, partners through greater profits, or technology providers through software subscriptions?
- 10
Are we at the end of legal AI as a software category and the beginning of AI as legal infrastructure?
For the last three years firms have asked: “Which AI tool should we buy?”
The emerging question is: “How do we weave AI into everything we do — and how do we get our own knowledge into the system we use, whichever one it is?”
How AI is Impacting You
If you haven't already, tell us how you're feeling about AI in the AI Survey (June 2026).
Where does your firm stand?
Take the AI Maturity Assessment