In a family business, the person who decides about AI should be the one who carries the consequence, not the most senior person or the most enthusiastic one. Split every AI decision into an owner's decision (data, money, reputation, who may use what) and a manager's decision (how a job gets done), write one name against each, and review the sheet twice a year.
That sounds dull until you picture the alternative, in a hypothetical family firm. Dad has a view at dinner on Sunday. A daughter back from university has already put the price list into a chatbot. The long-serving office manager has not been asked. Nobody did anything wrong, and nobody decided anything either. Below is a rule for splitting the decisions, a sheet you can copy to record them, and a worked example for a hypothetical 40-person firm.
Why do family firms adopt AI faster than they govern it?
Family firms are adopting AI faster than they are deciding who answers for it, and the research on larger family firms shows the gap clearly.
KPMG's 2026 US Family Business Report, published in August 2026, draws on 109 US family businesses in KPMG's Global Family Business Survey 2026. It found that 86% are piloting AI or have adopted it at scale, but only 40% have a formal AI governance framework. Half of them, 51%, named AI strategy and deployment as their biggest skills gap (KPMG, August 17, 2026). It is a small sample, so read it as directional rather than as a count of every family firm.
Deloitte Private's study, published on March 31, 2026, surveyed senior executives at 1,587 family businesses in 35 countries, all with revenues of at least US$100 million and a family holding a controlling stake. Fieldwork ran from March to June 2025. It found that 86% use AI actively or selectively, and 51% see inadequate technology adoption as a moderate or high risk to growth over the next 12 to 24 months. Its recommendations include prioritizing governance and risk management (Deloitte Private, March 31, 2026).
An earlier Deloitte survey of 100 leaders of US family-owned companies, with revenues of US$100 million to US$1 billion or more and fieldwork from April 28 to May 2, 2025, found that 42% said their board members have a limited understanding of emerging technologies (Deloitte, July 16, 2025).
Two cautions. The Deloitte firms are all large, with revenues of at least US$100 million, and the KPMG sample is small, so this is the pattern at the top of the family-business world, not a count of your neighbors. The reading is mine, not the researchers': when a business is run by people who share a surname and a kitchen, decisions get made in conversation, and conversation does not leave a record.
What is the Two Hats Rule?
The Two Hats Rule says every AI decision belongs to one of two hats, and the person deciding should be wearing the right one.
I call it the Two Hats Rule. It is my suggested way of thinking about decision rights, not an established standard:
The Two Hats Rule is how I suggest a family business decides who decides about AI: for every AI decision, name whether it is an owner's decision (risk, data, money and reputation) or a manager's decision (how a job gets done), write one name against it, and let nobody decide in the wrong hat.
Family businesses are the one place where people wear both hats all day without noticing the change. A founder who signs the checks is an owner. The same founder checking how invoices are chased is a manager. A son who is a shareholder is an owner; as head of sales he is a manager. When the hats blur, two things go wrong. A manager-sized choice, such as which tool the sales team tries, gets argued over at a family lunch. An owner-sized choice, such as whether customer files may go into a public chatbot, gets made by whoever happened to try it first.
The test for which hat applies is the consequence. If a mistake would touch customers' data, the firm's money or its name, it is an owner's decision. If a mistake would cost someone an afternoon, it is a manager's.
What goes on an AI decision rights sheet?
The sheet lists each AI decision, the hat it sits under, one named decider, who must be told, and when it is next reviewed.
Copy this table into a shared document and fill it in at one meeting. The rows are examples; add your own.
| Decision | Hat | Who decides (one name) | Who must be told | Review date |
|---|---|---|---|---|
| What customer or staff information may go into an AI tool, and what never may | Owner | [name] | Everyone, in writing | [date] |
| Which AI tools staff may use for work, and on whose account | Owner | [name] | Everyone | [date] |
| Whether customers are told when AI helped (see AI disclosure) | Owner | [name] | Customer-facing staff | [date] |
| Money spent on any AI tool or training | Owner | [name] | Finance lead | [date] |
| Which jobs the team tries AI on first | Manager | [name] | The team | [date] |
| Who checks AI output before a customer sees it | Manager | [name] | The checker, by name | [date] |
| What to do when AI gets something wrong | Owner | [name] | Everyone | [date] |
| Who answers a staff member's "may I use AI for this?" | Manager | [name] | The whole team | [date] |
Two rules keep the sheet honest. Every row carries a single name, never "the family" or "management". And the sheet records decisions already made: where a row is blank, nobody owns that choice yet, and that is the thing to fix. If you want the written policy that sits underneath these rows, start from the AI policy template a UK small business can copy.
What does it look like in a 40-person family firm?
It looks like four short conversations and one sheet, as this hypothetical shows.
Take a hypothetical family-owned builders' merchant with 40 staff. The founder, her brother and her daughter are the three shareholders. The daughter runs the trade counter and has been using AI to write quote follow-ups. The brother handles the money and distrusts all of it. The founder has not decided, so everyone assumes she is against it.
They sit down for an hour. Under the Two Hats Rule they sort the questions. The daughter's follow-up emails are a manager decision, and she may continue because she owns the result. Whether supplier price lists and customer accounts may go into a public tool is an owner decision, and the brother, who carries the money risk, decides it: no customer accounts, price lists allowed. Whether to tell customers that AI helped draft their quotes is an owner decision, and the founder takes it. Nothing moves faster, but nothing is argued about at the next family lunch either.
Notice what the sheet did. It did not settle who is right about AI. It settled who answers for what, which is the only thing a firm can act on.
What I Tell Business Owners
I tell owners that the next generation's enthusiasm is an asset to be given a job, not a decision to be argued over.
Across the business teams I train, the person who knows most about AI is rarely the person with the authority to decide about it. Younger family members and newer hires have the habits. Senior ones have the authority. If those two never meet in a written decision, the firm gets either a ban nobody follows or a free-for-all nobody owns. My advice is to give the enthusiast a defined manager's decision, with a name against it and a date to report back, and keep the owner's decisions with the people who carry the risk. If you want more on bringing a reluctant team along, how to get skeptical employees to use AI covers the people side, and shadow AI covers what to do when staff are already using tools you have not approved.
Succession adds a wrinkle. In the Deloitte US survey, 37% of leaders named a lack of next-generation interest as a hurdle to succession. That is a separate problem. AI is a place to hand someone real responsibility early, and a written decision right is a concrete way to do it. It is my reading, not Deloitte's finding.
What if a new family member or hire knows more than you?
Treat their knowledge as a reason to give them a defined job, not a reason to hand them your decisions.
Someone who is fluent in AI should write the first draft of the sheet's manager rows and propose the answers to the owner rows. The owner still decides the owner rows. This is the same logic as in AI training for new hires: bring the newcomer's skill inside your rules rather than outside them.
What can you do this week?
You can have the first sheet written in one sitting, using these steps.
- Pick one hour and invite everyone whose name is on the shares or on a department, plus the person who runs day-to-day operations if they are not family.
- Copy the table above into a shared document and add any decisions your firm has already faced.
- Sort each row into owner or manager. If the family disagrees, ask who carries the consequence.
- Write one name against each row. Leave any row blank that nobody will take; that is your list of open decisions.
- Put two review dates in the calendar, six months apart, and show the finished sheet to every member of staff.
Where to go next
If you are not sure where your family business stands with AI before you fill in the sheet, the free Business AI Readiness Scorecard takes about five minutes and gives the family a shared starting point. And if you want your team to learn to use AI well once the decisions are written down, that is what my AI training for business teams is built for. As I put it on my business page: "The hard part of AI at work isn't the technology. It's helping people use it well, safely and every day. That's what an educator does."
Sources and further reading
- 2026 US Family Business Report, KPMG US, August 17, 2026
- Family Business Technology Transformation, 2026, Deloitte Private, March 31, 2026
- Deloitte Private survey reveals board governance, succession planning and AI adoption top the agenda for family-owned enterprises, Deloitte, July 16, 2025
Dan Fitzpatrick is The AI Educator: a Forbes contributor, international keynote speaker and bestselling author who has delivered 500+ keynotes in 30 countries.
