Crosby's model—charging $250 to $1,000 per contract instead of $3,000 via traditional billable hours—is not a curiosity. It is a warning. For UK legal services firms regulated under the SRA Code, this story reveals that the economic foundation of mid-market practice is cracking. A contract that takes three days of a senior associate's time at £350/hour costs the client £8,400. Crosby does the same work with AI agents and human oversight for one-third the price, in hours not days. This works because contract review is pattern-matching at scale—exactly what modern AI excels at. The firms feeling this pressure most acutely are the mid-market generalists: too large to ignore the cost, too small to absorb margin compression without restructuring.
What Crosby represents is the acceleration of a shift already underway. Harvey, Luminance, and others have spent three years proving that legal AI can handle document-heavy work. But Crosby has done something harder: it has proven that fixed-fee models powered by AI agents (not just AI search tools) can undercut incumbents on both price and speed. The pattern is clear: commoditised legal work—contracts, due diligence, document review—will move to AI-first providers. The billable hour survives only where legal work requires judgment that cannot yet be automated. For most mid-market firms, the uncomfortable truth is that their current work mix sits somewhere in the middle, vulnerable on both flanks. Compliance with FCA Consumer Duty PS22/9 demands firms justify their fees. The billable hour, increasingly, cannot.
Trovix's view is direct: AI agents without proper human integration and explainability fail in regulated environments. Crosby's model works because it pairs AI with human oversight—but it does not solve the governance problem. UK law firms, insurers, and financial services firms must go further. They need AI that is integrated into their workflow (not bolted on top of it), that explains its reasoning to fee-earners and to clients, and that respects regulatory guardrails built in from the start. Generic AI tools like Microsoft Copilot are too blunt; specialist legal AI like Harvey and Luminance are powerful but are often deployed as search layers rather than workflow replacements. Trovix Brief was built for this exact problem: automating intake and matter initiation while keeping humans in control and maintaining an audit trail that satisfies the SRA and FRC. The difference is not subtle. A tool that extracts data and flags it for review is compliant. A tool that makes decisions without showing its work is not.
What should a mid-market law firm, insurance broker, or accountancy practice do on Monday morning? First, audit your own work mix. What percentage of your billable revenue comes from work that Crosby or similar competitors could undercut? Second, do not respond by defending the billable hour—respond by fixing your cost base and retraining your teams. Firms that move first to fixed-fee or value-based models backed by AI automation will capture clients fleeing incumbents. Third, invest in AI systems designed for your sector and your regulators, not generic consumer tools. The SRA expects firms to understand their AI systems. If you cannot explain how your AI made a decision, you have not implemented it properly. Fourth, recognise that speed and cost transparency are now table stakes, not differentiators. Your competitive edge now lies in the judgment layer: strategy, negotiation, advice. Build AI that handles the commoditised work so your people can do the work that justifies fees.
Source: Forbes