Crosby's model—AI agents plus humans reviewing contracts in hours instead of weeks, priced per contract not per hour—has caught Forbes' attention and venture capital's cheques. For UK mid-market legal practices, this story lands like a warning. The speed is real. The problem is what you lose chasing it. Crosby handles NDAs and services agreements for 100 companies; it does not navigate the SRA Code of Conduct for Solicitors, the FCA Consumer Duty (PS22/9), or the liability exposure when an AI agent misses a material term in a £2m commercial contract. Speed without governance is not innovation. It is risk compression.

This is the third wave of legal AI hype, and the pattern is consistent. First came document intelligence tools—Harvey, Luminance, Legora—which proved good at spotting patterns in text but created opacity: lawyers could not always explain why the AI flagged something. Then came the billable-hour disruptors, positioning themselves as pure cost-cutters. Now comes Crosby, which is essentially the same story with a freshly painted narrative: we are not just cheaper, we are faster. None of these models have solved the actual problem: how does a regulated firm prove to the SRA, the FCA, and the ICO that its AI system is compliant, auditable, and genuinely better than the human alternative? The EU AI Act and ISO 42001 are now raising the compliance bar everywhere in Europe. British firms cannot hide behind the old billable-hour shield much longer.

Here is what Trovix sees that this story does not say out loud. Speed kills trust in regulated practice. If your contract review takes three hours instead of three weeks, your clients do not celebrate—they worry. Did you miss something? Is the system cutting corners? The real lever is not velocity. It is rigour. A mid-market law firm should ask: does this AI system produce an audit trail that satisfies the SRA's expectations of professional competence? Can I explain every decision to a client or a regulator? Can I measure its error rate and its drift over time? Harvey and Luminance publish some benchmarks; Crosby does not. The firms winning this moment are not the fastest. They are the ones building AI governance into their process from day one—with Trovix Audit or equivalent, capturing every decision, every override, every exception—so they can defend the system when the SRA knocks on the door or a client sues.

If you are a mid-market firm watching this story: do not chase Crosby's speed. Instead, identify which contract types—NDAs, data processing agreements, routine supply contracts—genuinely benefit from AI acceleration without regulatory or reputational risk. Pilot them. Measure quality, not just turnaround time. Build an AI governance framework before you scale. Use Trovix Brief to automate intake and routing, so you get the speed benefit from process design, not from cutting human review. And monitor the regulatory landscape—Trovix Watch exists precisely for this reason—because the rules around AI in legal practice are hardening, not softening. The firms that survive the next three years will be the ones that proved they could use AI safely, not just quickly.

Source: Forbes