Forbes reports that $2.1 billion has flowed into AI-enabled legal services, with new entrants building 'law firms' that combine AI document generation with licensed attorney review to undercut traditional pricing. This story matters to mid-market UK regulated firms because it represents a false choice being sold to the market: either adopt AI aggressively and cut costs, or watch competitors do the same. The real issue is not whether these new services work—some do, technically—but how they handle the regulatory obligations that actually matter. The SRA Code (2019) and FCA Consumer Duty PS22/9 require that firms maintain explainability, accountability and auditability over decisions affecting clients. Most of the $2.1 billion is going into products that generate documents fast, not into systems that prove compliance with those obligations when the regulator asks.

This story is part of a larger pattern: the legal AI market is experiencing a funding-driven race to the bottom on cost, not a disciplined migration toward better client outcomes. We see the same pattern in insurance (where third-party AI claims assessment tools lack proper loss control auditing) and accountancy (where document processing AI lacks full chain-of-custody traceability). The assumption baked into nearly every venture-backed legal AI product is that speed and cost reduction are the primary selling points. But for regulated mid-market firms, they are not. Compliance, liability management, and the ability to defend decisions to regulators are. When Harvey or Luminance focus heavily on document assembly, or when Copilot for Microsoft 365 offers legal templates to any user without governance oversight, they solve a speed problem but create an accountability vacuum. The firms buying these products are exposed.

Here is Trovix's honest view: AI document tools and AI attorney review are useful but incomplete. A regulated firm cannot simply buy a document generator and call itself compliant. It needs three things that most legal AI vendors ignore. First, it needs an audit trail—not a log, a defensible record of what AI did, why, and which human reviewed it, that stands up to SRA examination or FCA investigation. Second, it needs governance that actually works: who can use this AI, for which matters, with what review gates, and how do you know when the AI has drifted off-policy? Third, it needs to understand its own liability exposure—what happens when the AI-assisted document creates a downstream problem, and who is responsible? Trovix Audit exists because we saw firms buying document AI and realizing too late that they had no way to prove compliance. The SRA's 2024 tech guidance and the emerging EU AI Act baseline for high-risk legal services both point to the same conclusion: automation without governance is regulatory theatre, not legal practice.

What should a mid-market law firm, insurance broker, or accountancy practice actually do right now? First, do not assume that lower cost means lower risk. A document assembly tool that costs 50% less but creates a 100% compliance problem is not an investment, it is a bet against your regulator. Second, if you are evaluating document AI—and you should be, the efficiency gains are real—make it conditional on implementing proper governance alongside it. That means defining which matters are suitable for AI assistance, which humans review the output before it reaches a client, and how you audit that decision-making. Third, treat regulatory change as an input to AI governance, not an afterthought. Trovix Watch monitors FCA, SRA, FRC and PRA guidance so you know when your AI approach needs to change—because it will, repeatedly, over the next three years. The firms that win in this market will not be the ones with the cheapest documents. They will be the ones that can prove they made the right decision, with the right oversight, every single time.

Source: Forbes