You know something you are not allowed to say out loud. The AI initiative needs the founder visibly in it, and you cannot tell them that without it landing as a criticism of the person who handed you the job. You have the mandate. You have the budget. The one lever you are missing is the founder’s own attention, and it is the one lever you are not permitted to pull directly. So you carry on, hoping the rollout earns the engagement on its own. It rarely does.
This post is about what you can do instead. Not a script for managing your founder behind their back. A few honest, sensible moves for drawing a senior person into a shared goal without overstepping. I should be straight with you on one thing before we start. The broad evidence here is solid, but the specific pull-in tactics are borrowed from adjacent fields rather than tested in AI settings. I will flag that as we go.
The choice you’re facing
You are choosing between two ways of handling a founder who has gone hands-off. One is to wait, run the programme well, and hope the results pull them in. The other is to influence up deliberately, framing your asks around what the founder already wants. The first is safer, slower, and usually stalls. The second is the one worth learning.
The reason waiting tends to fail is not about your delivery. Change-management research has pointed for decades to visible, sustained sponsorship from the top as among the strongest predictors of whether a rollout sticks. When the founder is invisible, the people below read the silence accurately. They conclude that AI is a side project a specialist is sorting out, rather than how the business now works. Good delivery cannot fully compensate for that signal, which is why the influencing path is worth the discomfort.
When influencing up is the right move
Influencing up is the right move when you genuinely share the goal and your founder has gone quiet rather than actively blocking you. The test is whether you would be comfortable saying every step to their face. Framing an ask around exit value, offering to automate a task they hate, agreeing where each of you decides, all of that passes. It is normal work with a busy boss.
The founder’s reluctance usually has rational roots, so meet it with respect rather than frustration. Many founders file AI under IT, a domain where they feel a specialist is more legitimate than they are. Many feel exposed stepping into an arena where they are a novice relative to their own reports, and in an investor-backed business, admitting a learning gap can feel risky to their standing. They are also running at the edge of their bandwidth. Sensible people behave exactly this way under those pressures, and your moves should make engaging feel low-risk rather than like a confession.
When to hold back and let the work speak
Hold back from active influencing when the founder is not reluctant so much as genuinely trusting you, and the programme is moving. If you have clear decision rights, a visible mandate they reaffirm when asked, and adoption climbing, you do not need to engineer their attention. Pushing harder there can read as insecurity, and it spends capital you may want later for a tougher ask.
The harder judgement is telling trust apart from absence. The signal to watch is what happens at the edges of the programme, the moments where a decision needs the founder’s weight behind it. A founder who is trusting you will back you when you ask, quickly and in public. A founder who has stepped out entirely will be hard to reach, will defer the decision, or will let it drift without resolution. If you are getting the second pattern, the work will not speak loudly enough on its own, and the influencing moves below become worth the risk.
What it costs to get wrong
Getting this wrong is expensive in two ways, and the second is the one people miss. The obvious cost is a stalled rollout, where the programme loses momentum because nobody senior is seen to own the direction. The subtler cost is the power vacuum created when a founder hands off ownership verbally then keeps intervening on key calls, teaching the organisation that AI is still the founder’s decision.
This reverse-delegation dynamic is well documented outside AI, in academic leadership settings and in the broader literature on undermined authority, and it maps cleanly onto the founder-delegate relationship. The practical damage is that your team learns to wait for the founder rather than work with you, so decisions slow and the mandate you were given becomes hollow. There is a deeper cost too. If AI gets built to mirror the founder’s instincts without the underlying process being documented, the business ends up more dependent on the founder, not less, which is the opposite of what an exit-minded owner is paying for.
What to ask before you commit
Before you commit to drawing your founder in, have the decision-rights conversation early, while it still feels like setup rather than a complaint. Agree where the founder engages and where they delegate, what the literature calls structured autonomy. Name the calls that are theirs, the calls that are yours, and how you handle the rest. Doing this at the start beats reclaiming authority after they have started intervening.
With the decision rights agreed, the practical first moves are small and framed around the founder’s own goals. Offer to automate one task they do every single day, just for them, off to the side, so they feel the value rather than being told about it. The point of doing it privately is that the founder gets to be a beginner where nobody is watching, which removes the status risk that keeps many of them at arm’s length. Reframe the ask so becoming an effective AI user reads as growth and status, not as approving more things. The same involvement described as making the founder powerful lands very differently from the same involvement described as the founder needing to sign off more work.
Lead every conversation with what they want, a more valuable business and fewer things only they can do, never with the tool. For an owner thinking about an eventual exit, that goal has real teeth, because buyers commonly discount a founder-dependent business by thirty to forty per cent. Used well, AI becomes the reason to document how the founder makes their best calls, which reduces that dependency rather than baking it in deeper.
None of these are proven AI methods. They are reasoned moves borrowed from delegation psychology and influencing-up work, offered as sensible practice rather than tested playbook. On Monday, pick the one daily task and the one goal-framed sentence, and start there.
