Legal hygiene is the degree to which a company has its legal obligations in order: contracts that are correct, IP that is demonstrably owned by the business, data processing that complies with the rules that apply to it, and files that hold up in the event of a dispute or audit. A buyer weighs this because any shortcoming here becomes the buyer's problem after the transfer, no longer the seller's. What used to be a checklist a lawyer ticked off once has become an ongoing process as soon as AI helps draft, check or process legally relevant documents. A buyer then asks not only whether the contracts are correct, but also who or what drafted them, and whether there is oversight of that.
In practice there are a few places where this becomes visible. Contract management: is it established which version is valid, and is there a trail of who approved which change. Data processing: is it recorded which personal data ends up where, including when an AI application processes or summarises that data. IP and ownership: is it clear that work created with AI assistance is nonetheless demonstrably owned by the business and not by an external tool or supplier. And oversight: is there a documented step in which a human approves or rejects an AI-generated document, with a reason, before it goes out. If that step is missing, a buyer sees a process that is fast on paper but that in practice nobody has checked.
AI already takes over parts of legal work today: searching contracts for deviating clauses, summarising files, drafting standard documents. That already happens today in some companies and not yet in others, and the difference rarely lies in the sector but in how the work is organised. Where AI produces a first draft and a designated person approves or rejects it with a reason, a file emerges that was created faster but is no less verifiable. Where AI produces a document that goes out unchecked, a file emerges that was created faster and is less verifiable, and that is exactly the difference a buyer seeks to establish. The three categories that run through all this work -- AI takes over the task, AI works under oversight that approves or rejects, or it remains human work -- determine not only the speed of the process but also who remains liable if something goes wrong.
This touches on how staff are deployed and which tasks have been redistributed. Separate statutory requirements apply to that, and those are not addressed here: what an employer does with its staff is up to the employer.
You get a first indication by checking three things. One: can you show, within a day, the valid version and approval history of your ten most important contracts. Two: is there a documented moment at which a human checks AI-generated legal documents, and can that moment be found in a log or email, not only as an intention. Three: is it recorded who processes which personal data, including the AI applications used in doing so, and is that record up to date. If any of the three is missing, that is a concrete point a buyer will probe during due diligence, and one on which the company's value may be adjusted.
The amount of time this measurement takes depends on how many contracts, files and data processing activities there are and how scattered they are currently recorded. A company with centralised contract management measures this in hours; a company where contracts sit in separate mailboxes measures it in weeks.
Improvement usually does not lie in new legal knowledge but in documentation: which step is done by whom, what oversight applies to it, and where the evidence of that is kept. For AI-assisted documents, that means building in a fixed checkpoint, with a reason for every approval or rejection, so that the process is both faster and traceable. This touches on how dependent the company is on the owner personally to know and do this kind of thing, which is addressed in what does AI do to owner dependency, and on the question of whether the savings AI delivers are embedded in the process or hang on one person or supplier, which changes how earnings quality is weighed as described in what does AI do to earnings quality. The management team also plays a role here: a buyer wants to know who on the team is responsible for this oversight if the owner is no longer there, something addressed in why does a buyer look at the management team.
The underlying question -- which work in this company can genuinely be taken over by AI, and which work continues to require oversight -- is answered per task by the FTE TO AI work scan.
Legal hygiene is one of eight drivers that together determine what a buyer wants to pay for the company, and how these drivers relate to each other can be read in how is the value of my company determined. A free value check of eight short questions, one per driver, gives a first picture of which driver is currently weighing most heavily on the company's price. The full value scan, with maturity scores per driver, supporting evidence and a two-year calendar towards the exit moment, is under construction.