Can a law firm sell the thing that lets clients need it less?
Sixty per cent of the 100 UK law firms in HSBC’s annual strategy and investment survey of the sector said clients moving more work in-house was disrupting them, City AM reported on 15 October. A day earlier, the Financial Times revealed emails, accidentally sent around Freshfields, showing its global managing partner, Alan Mason, working on moving the firm’s AI business into a separate company that could take outside investors or float. The two stories describe one problem from opposite ends. Clients armed with AI are questioning what an hour of legal work is worth, and at least one magic circle firm is asking whether its AI is worth more as a company than as part of the practice.
Strategic Insight: The survey and the emails are not separate trends. If clients do more of the work with software, the hourly model is under strain; if a firm owns software clients want, that software may be worth more on its own. Our news desk covered the Freshfields emails and the HSBC survey as they broke. This piece asks what they mean together.
Why the hourly bill is where the pressure lands
City AM’s reading of the HSBC findings is that firms are struggling to price specialist work, traditionally charged by time spent, now that corporate clients handle more matters through internal legal teams with the help of AI tools. Helen Drayton, who runs Penningtons Manches Cooper, said some clients now turn up with “their own AI-generated strategies which the lawyer must spend time advising against.” Her question: “How do you decide how to charge for that?”
HSBC’s Victoria Ritchie, who leads the bank’s professional and business services coverage, told the paper that AI is “already challenging the traditional law firm operating model, accelerating changes in how legal services are delivered, priced and resourced.”
The tools leaking work are sometimes the firms’ own
The line that connects the two stories came from David Beech, who runs the listed firm Knights. He said clients’ general counsel “may even be using the likes of AI solutions that partner firms have themselves recommended or developed to meet their needs.” The consequence, in his words: “This finds firms more pressured to make the case for an ongoing relationship.”
That is the context the Freshfields emails land in. A firm that builds good AI tools and hands them to clients is, on Beech’s account, already supplying some of the means by which clients need it less. Selling those tools through a separately capitalised company would turn that leak into revenue.
| Metric | Value | Strategic implication |
|---|---|---|
| Firms citing clients moving more work in-house | 60% (HSBC survey of 100 UK firms) | Demand for hours is being questioned by the people who pay for them |
| Firms citing technology in legal delivery as driving disruption | Almost three-quarters | Most of the sample sees technology changing how legal work is delivered |
| Firms calling AI integration a “major organisational challenge” | 73%, the challenge cited most often | Firms are still working out how to use AI in-house while clients adopt it |
| Revenue multiples Mason cited for legal tech | 25 to 40 times | The valuation case rests on software multiples, not practice economics |
Strategic Reality: Almost every firm in HSBC’s sample has a strategy for generative AI, according to City AM. The barriers firms named were time to explore the market, legacy systems and data, cost, and doubt about whether the spending will pay back. Having a strategy has not settled the return.
What the Freshfields emails propose
Writing to Ethan Klingsberg, who co-leads Freshfields’ US corporate and M&A practice, Mason said he was putting “a lot of work” into turning “our AI business in[to] a separate legal entity we can get external investors in / IPO”, the FT reported. He put legal tech valuations at between 25 and 40 times revenue. Two Wall Street banks, he wrote, “have reached out on this”. Klingsberg thought a standalone business would be “probably worth billions” and wanted to move fast, warning that “soon others will catch up and multiples will go down”. The FT set those claims against Harvey ($15.5bn) and Legora ($5.6bn).
Klingsberg’s sketch, as the FT described it, began with a separate company able to attract outside money and, later, to pass some back to the partners and the legal practice “in a tax efficient manner”. After that would come customers: “some large corporations with large legal departments”. Investors would be “sophisticated” ones, which he said “probably” excluded “Middle Eastern sovereign wealth money”. The thread reached far more people than intended when Mason forwarded it over a weekend to a large internal mailing list called “$Freshfields All”. It was recalled quickly, but not before some recipients captured screenshots.
Speaking to the FT, Klingsberg was more cautious. The idea had “just hit me last week”, the talks were “very early stage” and might come to nothing, and no tools had yet been sold to rival firms or to in-house legal teams elsewhere, the two buyer groups he named. He also gave the logic in a sentence: if a firm has built something valuable, “you don’t want to dilute its value by burying it in something else.”
Critical Context: The unit in question is Freshfields Lab, which the FT describes as building AI tools used by Freshfields’ own lawyers and by its clients, covering work from document analysis to litigation management and due diligence. Non-Billable reports it is co-headed by Gerrit Beckhaus and Lukas Treichl. Freshfields also agreed a deal with Anthropic this year on specialist legal AI that might eventually be offered to other firms.
What earlier firm spin-outs were reportedly worth
Other firms have separated a technology business from the practice before. Non-Billable lists three cases. In 2023 legacy Allen & Overy sold most of aosphere, its legal software arm, to Inflexion at a reported value above £200 million. Wilson Sonsini sold SixFifty last year, reportedly for up to $85 million. Travers Smith made Jylo, previously its legal AI unit, an independent software business in 2024.
Those are the reported figures. Harvey and Legora, the FT’s comparators, are valued at $15.5bn and $5.6bn. Legal Cheek notes a difference in kind too: Harvey and Legora each serve over 1,000 customers and run as independent technology companies, which Freshfields Lab does not.
Reality Check: Mason’s argument rested on the multiples of legal tech companies in general, and Klingsberg claimed Freshfields was “better than Harvey and Legora”. Neither is reported comparing the unit with earlier firm spin-outs. So far the valuation case is the two partners’ own, made in an internal email exchange, and Klingsberg describes the talks as at a “very early stage”.
What would decide whether the asset is real
Four questions follow from the coverage, and a carve-out would have to answer each:
- Value outside the firm’s own data. Klingsberg himself called it a “big question” whether outside buyers would still value the tools once they ran on their own data instead of Freshfields’. A tool that works because of the firm’s matters may not travel.
- A customer base of its own. The comparables have over 1,000 customers each, on Legal Cheek’s figures. Klingsberg pointed to one of the biggest FTSE 100 groups as the kind of buyer he had in mind and claimed “We’re better than Harvey and Legora!”, but no outside sale has happened yet.
- Timing. Klingsberg’s case for speed was that rivals would catch up and multiples would fall, which implies the valuation window is not permanent.
- A structure partners accept. The model discussed would eventually return money to the partners and the practice, so the partnership’s own economics are part of the design.
Who feels the pressure
Between them, the two stories touch four groups, from partners to the people who buy legal services.
| Group | What changes for them | What is still unknown |
|---|---|---|
| Partners at large firms | A separate AI entity could pass money back to partners and the practice (the FT, on Klingsberg’s plan) | Whether outside investors would accept a unit whose value depends on the firm’s data |
| UK firms in HSBC’s sample | 73% call AI integration a major challenge; barriers include cost and uncertain return | Whether pricing moves away from the hour, and how |
| General counsel and in-house teams | Some may be using AI tools their law firms recommended or developed (Beech) | Whether that changes how fees are negotiated |
| Investors and banks | Two Wall Street banks “have reached out”, Mason wrote (FT) | Whether a firm-born unit can earn software multiples |
The FT adds a qualification that matters here: AI is not yet good enough to disturb the market’s top end, so firms are pointing it at junior and routine work like drafting contracts and due diligence. We looked at the gap between firms that build AI and firms that buy it in our analysis of the build-or-buy choice for UK firms, which opened with Kirkland & Ellis setting aside $500 million for its own AI tools.
Competitive Reality: Others are attacking the hourly model directly rather than spinning tools out of it. Legal Cheek points to former A&O leader Wim Dejonghe, who has co-founded an AI-powered law firm focused on commercial contract review, a move it presents as a direct challenge to the billable hour.
What this means for general counsel
On Beech’s account, some general counsel may already hold AI tools their firms supplied. If clients are already using AI their law firms recommended or built, and firms feel “more pressured to make the case for an ongoing relationship”, the case for continuing to instruct a firm has to be made against what the client can now do alone. Should Freshfields sell its tools through a separate company, its clients could end up as customers of the software arm and the practice at once, which is a new kind of relationship for both sides to price.
The open problems
Data that may not travel
The asset Freshfields is weighing may draw part of its value from the firm’s own work. Klingsberg acknowledged as much when he told the FT that whether the tools would hold their value for outside buyers using their own data was a “big question”. A spin-out would have to show that its products work for buyers who do not have the parent’s matters behind them.
Pricing work the client has half done
Drayton posed her question without answering it. When a client arrives with an AI-generated strategy, the lawyer’s time goes on advising against it, and her question implies that time does not fit neatly into a bill for work done. HSBC’s respondents also named doubt about the return on investment as a barrier, so firms are unsure of both sides of the equation.
Selling to the people you bill
If firm-built tools help clients take work in-house, as Beech suggests some already do, a firm that sells its tools more widely may reduce demand for its own hours. A separate company could keep that tension off the practice’s books, but it does not remove it.
A window that may close
Klingsberg’s argument for speed assumes today’s multiples will not last. If he is right, any firm deciding slowly may find the comparables have moved before it reaches investors.
⚠️ Warning: The FT’s account rests on an internal email exchange and Klingsberg’s interview, in which he said the talks might come to nothing. Treat it as a signal of how senior partners are thinking, not as a decision.
What to watch
Three checkable developments will show which way this goes:
- Whether Freshfields formalises anything. A named entity, an appointed bank or an outside investor would show the emails were a plan. Silence would fit Klingsberg’s description of an idea that might go nowhere.
- HSBC’s next annual survey. If the share of firms citing clients moving work in-house rises from 60%, the fee pressure is spreading. A fall would suggest the 2026 figure was a peak of concern rather than a trend.
- The next firm-born spin-out and its price. Another UK firm following Travers Smith’s route with Jylo, at a valuation closer to aosphere’s reported figure than to Harvey’s, would indicate the market prices firm-born units as practice tools, not software companies.
Sam Elkin, finance director at Roythornes, told City AM that firms are wrestling with “what a fresh wave of potentially transformational technology will now mean for the business model of a law firm”. The Freshfields emails are one firm’s answer to that question, made in private and leaked after an accidental send. Firms without an AI unit to sell are left with the other half of the problem: pricing the hour when the client has already done some of the work.
Sources: Freshfields partners explore carve-out of AI unit worth ‘billions’, leaked emails reveal, by Olaf Storbeck, Suzi Ring and Kaye Wiggins for the Financial Times, 14 October 2026; and Law firms’ billing model faces disruption as clients turn to AI DIY advice (City AM, 15 October 2026), reporting HSBC’s survey of UK law firms. Earlier spin-out deals and Freshfields Lab’s leadership are from Non-Billable; customer numbers for Harvey and Legora, and the Kirkland and Dejonghe examples, are from Legal Cheek. We could not locate HSBC’s own release of the survey.
This strategic analysis was written by Resultsense, a UK-focused AI news and analysis publication. We will be watching whether any UK firm puts a law-firm AI unit in front of outside investors, and what price the market puts on it. Read more analysis at Insights, or get in touch.