Two companies, one playbook
ARR (latest)
Latest valuation
Total raised
Employees
Harvey San Francisco, founded 2022
Founded by Winston Weinberg (a former litigator) and Gabe Pereyra (a former DeepMind and Meta researcher), with OpenAI's Startup Fund as its first backer. Product: Assistant, Vault (document sets), Workflows and Agents, plus Knowledge, Contract Intelligence and a mobile app. Claims 80% of the Am Law 100 and five of the Fortune 10 as customers. In August 2026 it released Tenet, its own model, post-trained on Moonshot's open-weight Kimi K3 with Fireworks AI.
Legora Stockholm and New York, founded 2023
Founded by Max Junestrand, Sigge Labor and August Erséus. Its first client, Mannheimer Swartling, signed within a month, and the team worked inside that firm's office for nine months. Product: an "agentic operating system for legal work": Agent, Tabular Review, Legal Research and Workflows, with Word and Outlook add-ins and a white-label client portal. 130,000 lawyers a month across 2,100 firms and teams in 80+ countries; the US became its largest market in early 2026, and in-house teams are 40%+ of new business.
How they have been financed
Harvey's valuation rose 22x in under three years; Legora's rose about 8x in eleven months. Switch the chart between valuation and cumulative capital raised, and hover any round.
Hover or tap a round.
| Date | Company | Round | Amount | Post-money | Led by |
|---|
Rounds, amounts and leads from Sacra and Contrary profiles and company announcements; early valuations are often undisclosed. Seed and Series A figures differ slightly between sources.
Revenue growth and what investors paid for it
Both are among the fastest enterprise software companies on record: Legora took 18 months to reach $100M ARR and under six more to double it. The valuations work out to roughly 40 to 55 times ARR at each latest round.
| Measure | Harvey | Legora |
|---|---|---|
| Valuation ÷ ARR at latest priced round | $15.5B ÷ $400M ≈ 39x | $5.55B ÷ ~$100M ≈ 55x |
| At Legora's reported next round | $8.5B ÷ $200M ≈ 43x | |
| ARR per employee | ≈ $465K | ≈ $285K |
Legora's 2024 and 2025 ARR points are Sacra estimates. ARR per employee uses approximate headcounts.
What the business actually is: software, people, and embedding
Underneath the AI story, both companies run a hybrid of three things. The software is an interface and workflow layer over rented frontier models (OpenAI, Anthropic, Google) plus retrieval over the firm's own documents. The people are former lawyers and "legal engineers" who sit with each firm to drive adoption; about 10% of Harvey's staff do this. The embedding is getting into the firm's document systems, Word and Outlook, knowledge bases and client portals, so usage becomes habit and switching becomes a migration.
Sales model
Top-down enterprise sales to managing partners and general counsels, then land and expand. Harvey's median account doubles its seats within 12 months. Contracts are annual, per seat, with minimums: reported list prices for Harvey start around $1,200 per lawyer per month (20-seat minimum), with large firms negotiating far lower, plus onboarding fees. Legora is estimated at a ~$95K average contract. Both say they will move toward usage-based and outcome-based pricing, because a per-seat price sits awkwardly on top of a metered model bill.
Try the unit economics
An illustrative seat. Move the sliders to see how model costs and lawyer-led support decide the gross margin. Defaults are rough assumptions, not company figures.
This is why Harvey built Tenet. One report says it cut inference costs about 3x by routing work across Claude Opus and cheaper models on Fireworks AI. A secondary source claims agent usage drove its gross margin negative in mid 2026 before Tenet. That claim is unverified, but the mechanism is real: heavy agentic use on a flat seat price can cost more than the seat brings in.
The pitch to investors, and how strong each part is
Tap a company to see how its claimed advantages rate. The ratings are this report's judgment from the public record.
The investor story both tell: legal is a ~$1 trillion services market priced by the hour, most of its work is reading and writing documents, and AI adoption still trails capability. Whoever becomes the system lawyers work in captures part of that spend, and the market leader can later charge for outcomes rather than seats.
What could break it
Pick a scenario to see which advantages hold and which weaken.
The frontier-lab question
Anthropic launched Claude for Legal in May 2026, with practice-area plugins, connectors and an open plugin repository; its February preview triggered a legal-tech sell-off. Harvey and Legora are both launch partners and competitors at once. OpenAI, Harvey's first investor, is moving into legal too. The defence is that labs sell a general engine while legal buyers pay for workflow, liability-grade accuracy, procurement trust and change management. The test is whether firms keep paying a premium once the general tool is "good enough" for most tasks.
Other risks
- Head-to-head pricing. Two well-funded players chasing the same Am Law accounts push discounts and implementation spend up.
- Accuracy and liability. Courts keep sanctioning AI hallucinations, which raises verification costs and slows any move to outcome pricing.
- Model provenance. Tenet sits on a Chinese open-weight base; some firms and governments may object regardless of where it runs.
- Benchmark claims. Harvey's own LAB benchmark is the industry yardstick; Google's RRSI paper (Sept 2026) shows self-improving systems can overfit it, so held-out results matter.
What to watch and ask
- Gross margin after Tenet, and the share of usage on owned versus rented models.
- Net revenue retention and seat utilization at the top 50 accounts, as Claude for Legal enters the same firms.
- Pricing mix: how much revenue moves from seats to usage or outcomes, and at what margin.
Sources
- TechCrunch, Harvey hits $15.5B valuation (Sept 9, 2026)
- Harvey, $550M at $15.5B announcement
- CNBC, Harvey at $11B (Mar 2026)
- Sacra, Harvey profile (funding, ARR, pricing, go-to-market)
- Sacra, Legora profile
- Contrary Research, Legora
- Legora, Series D announcement
- Dealroom, Legora doubles to $200M ARR
- Runtime Wire, Legora seeks $300M at $8.5B
- The Next Web, Harvey's Tenet model
- Artificial Lawyer, Claude for Legal launches
- AI Weekly, Harvey cuts inference costs 3x (secondary)
- RRSI paper, arXiv 2609.24972
Private-company figures come from press reports and research firms, not audited filings. Where sources disagree (for example Harvey's share of the Am Law 100: 50% per Sacra, 80% per Harvey's September 2026 announcement), the company's own latest statement is used and the gap noted.