The headline is correct but the worry is misplaced. In-house legal teams are using AI to handle more routine work, and law firms are responding with rate cuts and alternative billing structures. For UK mid-market practices, this is not a threat—it is clarification. The SRA Code of Conduct requires you to act in clients' best interests, and if your client can do something better and cheaper with AI, they should. What matters now is where you add irreplaceable value. The firms panicking about volume loss are the ones who never had margin beyond time-based leverage anyway. The real conversation—which Law.com's story does not quite capture—is about which firms have built proper governance and which have just bolted chat interfaces onto their document stores.

This story is one frame in a much larger picture. For three years, legal AI vendors have promised 'efficiency gains' while most implementations have been shallow: document review, contract spotting, timesheet analytics. The vendors know it. Harvey and Legora built retrieval-augmented generation systems because they understood that unanchored generative AI in legal work is professionally negligent. Luminance went deep on due diligence because that is where the liability stays containable. Meanwhile, thousands of firms have deployed basic Microsoft Copilot over their file systems and wondered why their risk and compliance teams started asking hard questions. In-house legal departments are now asking the same questions. They are not replacing law firms because AI told them to. They are replacing law firms because the law firms they hired did not know how to use AI responsibly.

This is where Trovix's approach differs fundamentally. We do not sell you an AI copilot and hope you stay compliant. Trovix Audit exists specifically because mid-market regulated firms must demonstrate to the FCA (Consumer Duty PS22/9), the SRA, the ICO and increasingly the FRC that their AI systems are auditable, proportionate and not introducing hidden liability. When in-house counsel decide to build internal AI capability, they are making a governance choice as much as a technology choice. They are asking: can we prove this is working? Can we show our board it is not amplifying bias? Can we evidence that fee-earners are not using it to cut corners on advice quality? Trovix Aria is built on that premise—knowledge assistance that leaves an audit trail, not a black box. That is why in-house teams are winning some of this work back from external counsel. Not because AI is magic. Because someone actually implemented it properly.

Here is what you should do Monday morning. First, stop thinking of in-house AI adoption as competition and start thinking of it as a client capability you should be helping to build. Second, audit what AI your fee-earners are actually using right now—not what your IT procurement team licensed, but what they are actually running on client work. If you cannot answer that question quickly, you have a governance gap that will cost you more than any in-house AI ever will. Third, if you have alternative fee arrangements with in-house counsel, ask them what AI models and workflows they are using internally. That conversation will tell you whether you are commoditised or whether you are still doing work that requires actual legal judgment. Trovix Sift helps firms answer that last question by building AI capability that is transparent enough to explain to a client why they still need you. Because they will. They always do. But only if you can show them what you are actually doing with the technology.

Source: Law.com