Most Founders Don't Have an AI Strategy

Author

Aaron Hayslip

Published

Jun 26, 2026

Read

6 min

Practices

Automate, Delegate

Everyone's telling you the same thing: delegate it to AI.

I run a business built on delegation. At FreedUp, the whole thesis is that you double down on the few things you're great at that no one else can do, and you give everything else away. So when people say "delegate it to AI," I understand the instinct.

But that instinct is wrong.

You can't delegate to AI.

Delegation means handing ownership to someone else - a person who takes responsibility, uses judgment, and tells you when something's off. AI does none of that. When you "use AI," you're really doing one of two things, and neither one is delegation.

What You're Actually Doing When You "Use AI"

1) Augmentation

AI helps you do your own work faster - whether that's synthesizing a document, drafting an email, generating a report, or just thinking out loud - and it's genuinely incredible. I use it constantly.

But notice what's actually happening: you're still doing the work; you've just got a power tool now. And a power tool doesn't change the fact that doing the wrong work faster is still doing the wrong work.

2) Automation

There are two kinds of automation:

1. Schedule-based automation (a cron job). Example: generates my brief every morning at 8am.
2. Trigger-based automation. Example: when this happens, do that.

No-code automations have existed for fifteen years, and Zapier was built on exactly this back in 2011. The only thing AI adds is that the automation can now make judgment calls while it runs. That's more powerful - but it's still just automation.

Here's the catch nobody mentions: for your personal workflow, you have to build the automation yourself, because you're the only one who can. It's your Gmail, your calendar, your notes, and you can't hand those logins to anyone else. So you become the builder, and the moment you're the builder you're also the maintainer.

More often than not, the time you spend maintaining the automation outweighs the time it ever gave back.

Why Smart People Do It Anyway

Two reasons:

• Novelty
• The flex of getting to say you built an AI thing

Here's a real example. Alex Lieberman, the founder of Morning Brew, posted about the most powerful Claude skill he'd built: a daily planner. It runs off two markdown files, one for his priorities and one for his tasks, and every morning it reads both, makes judgment calls about his schedule, and creates his calendar invites once he approves them. The skill runs across about fifteen files, and only he can run it.

But why? Like, why doesn't he just... plan his day?

He's manufactured a pile of process documents and tech debt to replace something a person can do in ten minutes by just looking at priorities and tasks and planning the day. That isn't leverage. It's novelty.

And the cost is real, on three fronts.

The Three Costs

1) Your Time

AI is a dopamine machine - the same mechanism as social media - where you put something in and you don't know what you'll get back, so you keep pulling the lever.

It's dangerously easy to lose a whole weekend building something that's all wow and moves zero needles.

2) Actual Money

People have lost billions chasing this. And if Opus is the most powerful model, why wouldn't you reach for it every time?

So everyone's tokenmaxing, racking up enormous AI bills for marginal gains.

3) Results That Mostly Aren't There

In a survey of roughly 6,000 executives, nearly 90% reported no measurable productivity impact from AI over three years, and McKinsey found that only about 6% of companies are capturing significant value from it.

The thing that set those companies apart wasn't better tools. It was that they redesigned the workflow before they ever touched the tools.

The Non-Sexy Strategy That Works

So how do you get the upside without getting sucked into all of that?

It starts with knowing what you're actually responsible for in your business, and then running everything through a simple gate.

Filter #1: Is AI Letting You Do Work You Have No Business Doing?

If it is, that's a red flag, not a win.

You're not a software engineer, so should you really be shipping software as a founder? What you should work on is the overlap of:

• what you're good at, and
• what's good for the business

If something is good for the business but you're not good at it, AI doesn't fix that. It just walks you into a rabbit hole you can't see the bottom of.

(Prototyping something is a fair exception. Running production code you can't read is not.)

Filter #2: Could Someone Else Do This Instead?

Most of the time, the answer is yes.

Email, scheduling, calendar, admin, client onboarding and more - a person can own all of it. The pushback is always that AI is cheaper than a person, but it isn't, because you'll spend as much time maintaining the AI version as you would delegating it to a human.

And if that somehow isn't true for you, then your delegation is broken.

So the default question is simple: does this belong with me, or can someone else do it?

If someone else can, delegate it.

Hand off the outcome and coach them into an SOP as they go. People ask why they'd build an SOP if they're just going to automate it later, but that's backwards, because the SOP is the prerequisite for the automation.

The reason companies actually get value from hiring "heads of AI" usually has nothing to do with AI. It's that someone finally took inventory of the workflow and designed it on purpose - and that's where the gains live.

You shouldn't be the one automating it anyway. When I want a daily brief on my inbox, I don't build it; I ask my assistant to do it manually first. She starts by doing it manually, then using AI to augment her work, then she can automate it altogether once it's fine tuned.

The Actual Ladder

And if it genuinely belongs with you, augment before you automate.

Here's the ladder I climb:

1. Write the SOP as you work. Do the process by hand until you know every step.
2. Augment it. Pull AI into the steps that drag.
3. Build the skill. Turn the SOP into a repeatable skill that I tell to run while I oversee it doing what I'd normally do, in my pattern.
4. Schedule or trigger it. Put it on a cron job that runs every morning while I check the output.
5. Hand it off. Eventually I may even be able to give the management of that automation to my assistant, and now it's finally delegated - to a person who actually owns it.

Notice the order. The automation comes last, and only after a human process has earned it.

Most things never make it past step one or two, and that's exactly the point.

So here's my actual flex, the one I'd put up against anyone's fifteen-file daily planner:

I have very few personal AI automations.

That isn't me being behind.

The Fix

Here's where this is all heading. FreedUp already gives founders the system: an executive assistant who owns the work, plus the SOPs and delegation built around them.

What we're adding next quarter is the layer that's been missing, an actual AI strategy that sits on top of all of it, so the augment-then-automate ladder I just walked you through gets built for you instead of by you.

Want to be first in line when it goes live?