The pitch has been the same for two years. A recruiter, or a software firm selling to recruiters, says the shortlist can now be built by a model, the outreach written overnight, and the candidate ranked against the job description before anyone has spoken to them. LinkedIn has put generative tools inside Recruiter. The applicant-tracking systems have followed. A number of search firms have announced a platform. Most of what sits underneath is a large language model writing a message, or matching keywords on a CV to keywords on a spec. It is useful. It is not a picture of the market.

The hype landed because the market got harder to read, not because the technology was ready. NALP’s 2025 survey, published in 2026, put US lateral hiring up 16.4 per cent on the year before. Partner hiring rose 17.8 per cent and associate hiring 17.1 per cent. Associates were still 58 per cent of lateral hires, partners 22 per cent. SurePoint counted 13,460 lateral hires in the Am Law 200, up 8.8 per cent, the highest since the post-pandemic spike. The NALP Foundation had associate attrition at 19 per cent, and 83 per cent of those who left had been at the firm under five years. London does not publish the equivalent. From the City mandates, it is the same shape: fewer speculative processes than 2021, and more partner moves that are real. Volume at that level punishes a firm that is reconstructing the market by hand.

What other firms are mostly doing with AI is speeding up the old method. A CV is screened. A note is drafted in the candidate’s practice area. A list is scored. Some of it is genuinely quicker, and a junior researcher who used to spend a day on a first pass now spends an hour. The shortlist at the end is still whoever was on the database and whoever replied. A hiring partner who has sat through three of these processes can tell. The emails are better written than they were in 2023. The names are not.

There is a second, smaller use that does matter, and it is not recruitment at all. Law firms are hiring lawyers who can work with the tools on the client side. SurePoint recorded AI-related lateral hiring in the Am Law 200 up 68 per cent in 2025, associate hiring in that pocket up 106 per cent. Firm Prospects logged 15,988 moves into and out of the Am Law 200 in the first half of 2026, of which 46 lawyers went to AI companies, Harvey taking 22 of them. That is a real destination. It is 0.3 per cent of the moves. It is not how searches are being run.

We took a different cut. HIA Legal has invested substantially in LexQuest.ai, a private candidate relationship workspace, built as market intelligence rather than as an outreach engine. It tracks lawyer movement inside the Am Law 200, the Magic Circle and the other top-tier firms we cover, in London and in the US: who moved, when, from which practice, and into what. The relationship history sits on top of that, so a search does not open on a cold list. We are still building it. A record of movement is never finished.
The practical difference is the start of the mandate. We are not waiting for replies to tell us where a group has gaps. A finance team that has hired three times in a year, or a private equity practice that has lost two counsel in a quarter, is visible before the first call. The longlist is usually shorter than the one a generalist firm sends. People are approached where the practice fit is already there, not because a model liked the wording of a CV. Whether someone will sit with a particular team is still a judgement. LexQuest.ai does not make it, and we would not want it to.

Retention is the part of the hype we think is half right. The attrition figures are a market fact. A client who is told that a band of seniority is moving across several firms, not only inside their own building, can have the conversation before the resignation. We are not an HR system. We are watching the same movement we use for search.
AI has changed the speed of a bad process, and it has given firms a new specialist to hire. It has not, on its own, told anyone who is actually available. We would rather be early on a move, and right about the practice, than first to send the message.