You run a triage on your own attention without thinking about it. There is a fifth of the business that genuinely needs your head in it, the things that move the valuation, win the accounts that matter, set the direction. Everything else gets handed to someone good. You learned that rule because your bandwidth ran out years ago and the business kept growing anyway. It is the right instinct, and it is most of what has kept you sane.
Then AI lands on the list, and the rule misfires. Some of it is obviously someone else’s job. Some of it, on closer reading, is yours in a way that does not feel technical at all. The trouble is that the part worth keeping looks, on the surface, exactly like the part worth handing off.
What is the founder’s 80/20 on AI?
It is the question of which slice of the AI mandate needs your personal attention and which is a clean handoff to a capable operator. The eighty percent is the operational programme, the tooling, the vendors, the rollout. The twenty percent is the work that changes what your business is worth, mainly using AI to reduce how much the business depends on you.
Many founders sort by what feels technical. The sort that matters is by value. The reason the instinct misfires is that your usual 80/20 sorts by competence and comfort. You keep what only you can do and hand off what someone else does better. With AI, comfort points you the wrong way. The tooling feels like the part you should learn, because it is the visible, novel thing. The dependency work feels delegable, because it looks like a documentation project. The values are reversed under the surface.
Why does the split matter for your business?
Because the two slices have different consequences, and getting them backwards is expensive in a way that does not show up for years. Hand off the operational eighty percent badly and you lose some time and money on tools that do not stick. Hand off the high-value twenty percent and you delegate away the one piece of AI work that lifts your exit multiple.
The cost is invisible until a buyer prices it in. The number underneath this is the founder-dependency discount. M&A advisers describe owner dependency as a leading discount to an exit multiple, with buyers commonly applying 30 to 40 percent reductions where operations, relationships and decisions sit with the founder rather than the system. A firm earning the same profit can be worth a third less because the earnings walk out of the door when you do.
AI is one of the few tools that can attack that discount directly, but only if you point it at the right work. Used to codify how decisions get made, it reduces dependency. Used as a productivity toy that stays in your hands, or built to mirror your instincts without documenting the reasoning behind them, it can make the business more dependent on you while looking modern. The split decides which of those you get.
Where will you actually meet the high-value twenty percent?
You meet it in the room where the team tries to document a decision that currently lives only in your head. The high-value work is sitting with the people who will use AI to capture the judgement that routes through you, the pricing calls, the client reads, the go or no-go instincts. That work needs you present, because the knowledge being captured is yours.
It is the work that moves a founder-led business towards one that runs without you. The rest of where you meet AI is genuinely the eighty percent. Choosing a tool, managing the vendor, integrating it with the systems you already run, training the team, measuring whether it stuck. That is a real programme and it takes real skill, but it is an operator’s job. A good head of operations or a fractional lead can run it better than you can, and your involvement there is the thing to delegate without a second thought.
The trap sits in the overlap. The dependency work looks like a documentation exercise, which is delegable, so founders hand it to the same operator running the tooling. The operator then captures process without the founder’s reasoning in the room, and the firm ends up with AI that automates the surface of your decisions while the actual judgement stays locked in your head. Apparent progress, worse exit readiness.
When do you keep AI on your desk and when do you hand it off?
You keep it when the work touches the exit multiple and hand it off when it touches the operation. That test cuts cleaner than asking how technical something feels. Codifying your judgement and the sponsorship signal that tells the team this is how the firm works now both stay with you. The tooling, the vendors and the rollout go to the operator. Sort by value to a future buyer.
There is one failure pattern worth naming, because founders fall into it after they have done the sort correctly. You delegate ownership of the programme, then keep intervening on the decisions inside it. You overrule the tool choice, you reverse the rollout sequence, you take back the calls. The research on reverse delegation is blunt about what this teaches an organisation, that the authority you handed over was never real. The delegate stalls in a power vacuum and the team learns that AI is still your call after all.
The clean version is the opposite. You hand the operator the operational eighty percent with genuine decision rights, you keep your head in the dependency work, and you provide the visible sponsorship that change-management research consistently identifies as one of the strongest predictors of whether a tool rollout sticks. Kyndryl found around 70 percent of leaders say their workforce is not ready, with only 14 percent having aligned their people, technology and growth goals. The sponsorship signal is how that alignment happens, and it is yours to give.
What related ideas are worth holding alongside this?
Three ideas sit close to this one and sharpen it. The first is the difference between delegation and abdication, the line between handing an operator real ownership and walking away from the leadership signal that makes adoption work. The second is the dependency paradox, where AI built to mirror the founder makes the business more reliant on them. The third is decision rights, the agreement about which calls belong to the founder.
Held together, these reframe what the AI mandate is for. Treated as a leadership and operations decision rather than a technology one, it becomes a tool for reducing the single largest discount a buyer will apply to the business, the discount for needing you. That reframe is what turns a vague modernisation brief into a sharp one, and it is the reframe that decides whether the AI work earns its place in your twenty percent or belongs in someone else’s eighty.
If you are sorting the AI mandate right now and the line between the two piles is not obvious, that is the conversation worth having before you hand anything over. Book a conversation and we will work out which slice is genuinely yours.
