Harvey just closed $200 million at an $11 billion valuation, backed by Sequoia, GIC, a16z and Kleiner Perkins. The startup has built $190 million in annual recurring revenue in under five years. For UK legal, insurance, financial services and accountancy firms, this is not a story to admire from a distance. It is a warning. A well-funded US legal AI platform has achieved scale and institutional credibility. It will now hunt for market share in the UK. And many mid-market firms will be tempted to adopt it wholesale, believing that offshore AI deployment solves their automation problem. It does not. It solves a different problem — the American one.

Harvey's valuation reflects a real trend: law firms worldwide are moving from document review automation into workflow intelligence. The next tier is not better document reading — it is understanding matter context, predicting outcomes, and automating decision routing. Harvey, Luminance, Legora and others are all chasing this same layer. But the industry is making a structural mistake. It is treating 'legal AI' as a single category when it is actually three separate problems: extraction (finding data in documents), reasoning (applying law to facts), and governance (staying compliant with FCA Consumer Duty PS22/9, SRA Code of Conduct, and emerging standards like ISO 42001). Most platforms solve the first two. None solve the third adequately because regulatory risk is local and contextual. A US platform trained on US case law and depositions does not understand how the SRA expects you to justify an AI decision to a client, nor how the FRC's ISA UK standards apply to your AI audit trail, nor how the ICO expects you to handle prompt injection attacks under UK GDPR.

Here is Trovix's honest view. Harvey's $11 billion valuation is deserved for what it does: it reads English language documents at scale and finds relevant passages. That is worth billions to law firms drowning in discovery. But it is not worth the cost of implementation for a UK mid-market firm if you bolt it on without governance, without regulatory mapping, and without a way to explain to your compliance function and your regulator why you chose it. The risk is not that Harvey is bad. The risk is that firms will treat it as a turnkey replacement for human judgment instead of as a classified tool within a larger AI operating model. Compare this to how Trovix approaches the same problem: we start with regulatory risk mapping (what does SRA say about this, what does FCA say about that), then we build extraction and reasoning on top of that foundation. We do not ask you to adopt a US legal AI and hope your UK compliance function figures it out. We ask you to design the compliance layer first, then choose the extraction engine. Trovix Watch monitors regulatory change so you know when to update your AI governance. Trovix Sift handles document extraction but within a governance framework you control. And Trovix Brief automates intake while maintaining the audit trail your regulator expects to see.

What should you do right now? Do not wait for Harvey to open a UK sales office. Do not assume that because a platform is venture-backed and American it is regulatory-safe for your firm. Instead, audit your current AI deployments (yes, ChatGPT counts) against FCA Consumer Duty PS22/9 and SRA Code section C4.3 on client information. Map what you actually need to automate: is it really document reading, or is it intake, or is it matter intelligence? Then decide whether you need a US platform (extract and reason) or a UK-governed system (extract, reason, and justify). The firms that win the next two years will not be the ones with the most sophisticated AI. They will be the ones with the most defensible AI — the ones that can show a regulator exactly why they chose what they chose and what they did to protect clients.

Source: TechCrunch