Crosby's AI law firm proves that flat-fee, agent-driven legal delivery is viable at scale. UK firms still building on legacy infrastructure and billable-hour incentives are now competing at a disadvantage that will only widen.
Legal Tech  Trovix ReachLegal Services

Crosby's operating model—30 lawyers plus AI agents reviewing commercial contracts in hours rather than weeks, at flat fees of £250–£1,000 instead of six-figure retainers—is not a novelty. It is a competitive price signal that mid-market UK law firms, insurers and professional services practices can no longer ignore. The firm is handling service agreements, data processing agreements and NDAs for 100 companies at velocity and cost that traditional billable-hour practices cannot match. For in-house legal teams at UK financial services firms already under FCA Consumer Duty pressures (PS22/9) and insurers managing Lloyd's Blueprint Two requirements, this represents a real alternative to their traditional counsel. The question is no longer whether AI will disrupt legal services. The question is which UK firms will lead that disruption and which will become margin casualties.

Crosby's success sits within a larger pattern: the death of billable hours as the default pricing model for commodified legal work. We have seen versions of this argument before—Harvey and Luminance both promised to transform document review through LLM-based systems. What distinguishes Crosby is not the technology itself; it is the business model. They have removed the friction between AI capability and client delivery by pricing flat, removing the economic incentive to bill for time rather than value. This breaks the incentive structure that has kept most law firms locked into incremental productivity gains. The SRA's Code of Conduct for Solicitors emphasizes outcome-focused regulation, not process prescription, which means firms have room to experiment. Yet most have not. Instead, they have layered generalist AI assistants—Microsoft Copilot, basic contract analysis tools—on top of unchanged billing mechanics. The result: faster billable hours, not better client economics.

Trovix's view is direct: flat-fee, AI-driven service delivery only works if your AI infrastructure is not bolted onto legacy systems—it must be integrated into intake, triage, drafting and governance from the ground up. Crosby appears to have done this. Many UK firms using generic LLM tools or point-solution contract analysis platforms have not. The deeper issue is that contract review at speed requires more than language models. It requires domain agents trained on your firm's precedent library, integrated with your matter management system, auditable for regulatory and compliance purposes (particularly important under the ICO's UK GDPR requirements and emerging EU AI Act compliance expectations), and capable of escalating exceptions to humans in a structured way. Trovix Brief automates intake and triage precisely because these steps are where velocity breaks down in practice—not in the review itself, but in deciding what gets reviewed and by whom. Without fixing intake, you cannot achieve Crosby's throughput, and without throughput economics, you cannot move away from billable hours. Copilot and Luminance are good tools; they are not sufficient to achieve this shift on their own.

If you are a mid-market law firm, insurance company, financial services practice or accountancy firm with in-house legal, act now on three fronts. First: audit which of your highest-volume, lowest-margin services could be priced flat without sacrificing quality. Commercial contract review is the obvious starting point, but many practices have equivalent volumes in conveyancing, compliance documentation, or regulatory filing preparation. Second: implement AI governance frameworks before you scale AI agents. The FRC's ISA UK auditing standards and emerging AI risk frameworks (PRA SS1/23 thinking) require transparency on how automated decision-making affects your regulatory obligations and client outcomes. Trovix Audit exists precisely because firms cannot scale flat-fee AI delivery without proving that AI decisions are defensible, traceable and compliant. Third: do not wait for your industry association or regulator to signal it is safe. Crosby did not ask permission; they built the model and proved it worked. Crosby will consolidate market share while competitors debate methodology.

Source: Forbes

Related Trovix product:

Trovix Reach →Book a demo →