Forbes reports that legal-tech startups attracted $2.1 billion in the first half of 2026, with a wave of companies now automating routine work like contract review and incorporation. For mid-market UK law firms, insurers, and accountancy practices, this matters because the competitive pressure is real. If rivals are using AI to handle document-heavy work faster and cheaper, standing still looks like commercial suicide. But here is the uncomfortable truth: capital flowing into legal AI tells us nothing about whether these firms can safely deploy it under the SRA Code, FCA Consumer Duty PS22/9, or the ICO's UK GDPR framework.
What this $2.1 billion story actually reveals is a widening gap between innovation speed and governance maturity. Products like Harvey, Legora, and Luminance are genuinely capable at document classification, contract abstraction, and risk flagging. Microsoft Copilot is pushing AI deeper into everyday workflows. All of them work. But the investment boom has created a dangerous assumption: that buying the tool solves the problem. It doesn't. The real bottleneck for mid-market firms is not access to AI—it is the ability to implement, audit, and justify it to regulators and clients. Too many practices are adopting these tools without documented AI governance frameworks, impact assessments, or explainability protocols. When the ICO or SRA comes calling, a pile of venture capital means nothing.
Our view at Trovix is that the $2 billion has largely gone to building better models and prettier interfaces, not to solving the governance layer that actually matters for regulated firms. This is a market failure. A mid-market practice cannot just drop Harvey into contract review and hope for compliance. It needs to know what the AI is doing, why it made that decision, what it got wrong, and how that decision affects the client. It needs to monitor for model drift, bias, and regulatory change. It needs an audit trail. Most of the unicorns in that $2.1 billion are not building governance-first—they are building capability-first and treating compliance as a checkbox. Trovix Audit exists because that checkbox approach fails. We build from the assumption that if you cannot explain and justify your AI decision under FCA or SRA scrutiny, you should not make it in the first place.
If you are running a legal, insurance, financial services, or accountancy firm with 50 to 500 people, here is what you should do right now. First, stop waiting for a 'trusted' AI vendor to solve this for you. Second, audit which of your current document workflows actually need automation and which ones just feel like they should. Third, before you deploy any tool—Harvey, Luminance, Copilot, or anything else—build a governance framework that covers model selection, data handling, explainability, and monitoring. Trovix Watch will keep you current on regulatory expectations as they evolve. Fourth, demand that your AI vendor—or your internal team—can produce evidence of what the system did, why it did it, and how it performed. If they cannot, do not buy it. The firms that win over the next 18 months will not be the ones with the fanciest models. They will be the ones with the strongest governance. The $2.1 billion is being spent on the wrong problem.
Source: Forbes