Legal hygiene means: are the contracts, IP rights, employment relationships and licences documented in such a way that a buyer, after the acquisition, does not still have to negotiate over things he thought he had already bought. For a business without AI use, that is a matter of checking contracts, reviewing files, done. For a business where AI is part of the work, a layer is added: who owns what a model has produced, on what data it was trained, and which agreements with suppliers and customers cover that.
That layer is not theoretical. A quote drafted by a language model, a customer analysis that comes out of an AI tool, a contract clause that was generated automatically: if nothing has been recorded about this, a buyer buys uncertainty along with it. Not necessarily because something is wrong, but because nobody has checked it.
AI takes over work in three categories that run through almost every part of a business: tasks a system can handle entirely, tasks where a human approves or rejects with reason, and tasks that remain human work. Legal work often sits in the middle. Contract analysis and risk flagging can be partly automated, but the assessment of whether a risk is acceptable remains with a human.
That middle segment is exactly where hygiene becomes vulnerable. A business that uses AI to scan contracts without recording who reviews the outcome leaves a gap that a buyer will notice during due diligence. A business that does the same thing but with a documented oversight process, with a name and a reason behind every rejection, shows that the system is not a black box but an extension of a decision framework that is transferable.
The difference between those two businesses is not whether they use AI, but whether they have documented what AI did and who bears responsibility for it. That documentation is precisely what legal hygiene is about today.
A buyer does not look for a statement that the business uses AI. He looks for traces that show the use is controlled:
If that record is missing, a buyer does not price in the risk that something is wrong, but the risk that he cannot rule it out. That is more expensive.
A first picture emerges by going through three questions, per process where AI plays a role.
One: is it documented who is responsible for what when AI partly takes over a task. Not as an organisational chart, but as a document that would convince an outside party.
Two: do the contracts with customers, suppliers and software partners align with what actually happens. A contract that still assumes fully human work while practice has already changed is a gap that surfaces at transfer.
Three: is there a record, even an informal one, in which assessments of AI output are logged. Not to prove that AI performs well, but to prove that someone checks it.
Anyone who wants to show this for their own business, task by task, will find the underlying question — which work here can genuinely be taken over by AI and which work remains human work — answered in the work scan by FTE TO AI, task by task.
Legal hygiene does not stand alone. A saving that depends on a single AI supplier without contractual coverage weighs differently in how you assess profit quality in a business where AI is part of the work than a saving that is anchored in the process itself. A customer relationship that runs through an AI tool without the ownership question being settled also affects how you measure customer concentration in a business where AI is part of the work. And if oversight of AI output rests solely with the owner, that counts towards how you measure how much runs through the owner: hygiene that has not been transferred to the team is not hygiene that is sold along with the business.
If this topic touches on personnel decisions — for example because work is shifting from human to system — separate statutory requirements apply; this page does not offer advice on that.
The full value scan maps eight drivers, each with a maturity score and evidence, plus the owner-dependency index and a two-year calendar towards the exit moment. That scan is under construction. Available now is the free value check: eight short questions, one per driver, giving a picture of which driver is weighing most heavily on your price today. Anyone who then wants to know when preparation for a sale should begin will find that worked out in when you should start preparing a sale.