Crosby's model is undeniably attractive on paper: use AI agents to do initial contract review, plug the gaps with qualified lawyers, charge $250–$1,000 per contract instead of hourly rates, deliver results in hours not weeks. Forbes calls it buzzy. Startup founders love it. But here is what the story does not ask: who is liable when the AI misses something material? What happens to the SRA Code of Conduct when speed becomes the primary metric? And how does a firm defend its AI decisions to an FCA investigator or a client claiming breach of duty? For mid-market UK legal, insurance and financial services firms, this story is not really about speed. It is about what happens when you optimise for throughput without building the governance structure underneath.

This news belongs to a broader pattern we have watched for three years now. Generalist AI tools (Claude, ChatGPT, Copilot) flooded the market. Specialist legal AI followed (Harvey, Luminance, Legora). And now pure-speed players like Crosby are proving the market will reward whoever gets a document out the door fastest, regardless of what the SRA or FCA thinks about oversight. The same is true in insurance and accounting. The firms winning right now are not the ones with the best AI — they are the ones brave (or reckless) enough to skip the boring part: governance. That works until it does not. And when it does not, it usually happens in front of a regulator.

Trovix's view is simple: speed without control is just liability on a timer. Crosby's approach works because it leverages human lawyers at the end of the chain — the human check is real. But that model breaks at scale. If you are reviewing 50 contracts a day, the human lawyer becomes a bottleneck. You either hire more lawyers (killing your margin) or you rush the review (killing your compliance posture). The firms actually winning long-term are the ones building AI workflows with built-in governance from the start. That means Trovix Audit — a compliance dashboard that logs what the AI decided, why it decided it, and who signed off. It means Trovix Sift for reliable data extraction instead of hallucination risk. And it means Trovix Aria so your fee-earners can explain the AI output to clients and regulators with confidence. Crosby's model is optimised for venture funding rounds. The UK regulated model should be optimised for surviving a regulatory review.

If you are running a mid-market legal, insurance, financial services or accountancy firm, the action is not to copy Crosby. It is to ask yourself: do we have an AI governance trail? Can we defend what our AI did in a breach inquiry? Do our lawyers understand what the machine actually did, or are they just rubber-stamping it? If the answer is no to any of those, then speed is killing you, not helping you. Start with governance. Build the audit trail first. Then accelerate. That order matters.

Source: Forbes