From Disney to solo: what 15+ years inside enterprise taught me about AI

I spent 15+ years inside enterprise. Walt Disney World, NCR, Wyndham Worldwide. Big systems, big teams, big stakes, big timelines.

Then I left to build solo.

Most of what I read about “AI in the enterprise” online sounds nothing like what enterprise actually is. The takes are either too rosy (everything is being transformed!) or too dark (everyone is being replaced!). The truth is messier, slower, and more interesting.

Here’s what I actually learned in the rooms — and how it shows up in what I’m building now.

Most “AI projects” aren’t AI projects

When a Fortune-class company says they’re “doing an AI project,” what they usually mean is they’re doing an org-design project with an AI sticker on top. The model isn’t the bottleneck. The handoffs are.

Who owns the dataset? Who reviews the output? What happens when the model is wrong? Who has authority to override it? How does the audit trail work? What’s the SLA when it goes down? Which existing team is being reshaped by this — and who told them?

The model picks itself in 30 minutes. The rest is six months of meetings.

This is why the AI demos that wow people in conference keynotes often die quietly inside the company. The demo solved the easy part. The hard part is everything that happens after the demo.

The slowest part isn’t the technology

I watched genuinely useful AI workflows take 8+ months to land in production at one of those brands. Not because the tech was hard. Because:

• Legal review (ownership, liability, training data, output rights)
• Procurement cycle (vendor evaluation, contract negotiation, budget)
• Security signoff (data flow, access controls, audit logs)
• Change management (training the team, updating SOPs, comms to stakeholders)
• Political negotiation (whose budget owns it, whose KPI improves, whose role is affected)

AI doesn’t shorten any of those. If anything, it lengthens them, because each is now arguing about more things — including whether AI is even the right answer.

For solo operators reading this: this is your unfair advantage. You can run the same workflow in three days that takes enterprise three quarters. Don’t waste it.

The fastest wins aren’t the moonshots

The AI workflows that landed at the brands I worked at weren’t the splashy ones. They were the boring ones:

• Workflows that already had clean inputs and clean outputs
• Workflows where one specific person already hated doing them
• Workflows where the cost of being wrong was low and easily caught
• Workflows where the ROI was visible in week one, not quarter four

These ate overnight. The fancy ones — “let’s build an AI assistant for our entire customer service org” — are mostly still in pilot.

Lesson for builders: the same pattern works for your own work. Find the weekly thing you hate that has clean inputs and clean outputs. AI eats that first. Build outward from there.

The biggest blocker is rarely capability

Will the C-suite trust a model with sensitive customer data? Will the customer trust the output enough to act on it? Will the team trust the AI recommendation when it conflicts with their experience?

These are human problems, not model problems. They get solved with transparency, with audit trails, with calibrated uncertainty, with humans in the loop, with reputational risk frameworks. Not with a better model.

The brands that got AI to actually move the needle were the ones who invested in the trust infrastructure as seriously as they invested in the models. The ones that didn’t are still in pilot purgatory.

The “AI replacing jobs” narrative is mostly wrong inside enterprise

Inside the rooms I sat in, AI wasn’t replacing jobs. It was replacing tasks inside jobs. The jobs are getting reshaped — sometimes radically — but they’re not being deleted in the way the headlines imply.

The people getting left behind aren’t the ones whose jobs got “automated.” They’re the ones who didn’t reshape with the change. The role is still there. The shape changed. The people who couldn’t see the new shape — or didn’t want to — got quietly worked around until their role became indefensible.

This is the part I worry about most for the people I care about. Not “AI is taking jobs.” It’s “the work is changing and the systems for noticing aren’t keeping up.”

How this shows up in Waymaker

Everything I learned inside enterprise shaped what Waymaker is — and what it deliberately isn’t:

It’s not another AI tool. The market doesn’t need another model wrapper. It needs an operating layer that holds the plan, captures the patterns, and watches the gaps. That’s the seat I’m building for.

It’s built for solo and small-team operators. The people who don’t have a 6-month procurement cycle. The ones who can ship in a week what enterprise will take a year to approve. Your speed advantage is real — Waymaker is the system that makes you not waste it.

It’s honest about the gaps. UX, security, judgment, trust — these are the gaps AI can’t close on its own. Cofounder is built to watch them, the way the security and design and change-management teams watched them inside enterprise.

It captures what works as reusable plays. Inside enterprise, every project produced an artifact: a postmortem, a runbook, a lessons-learned doc. Solo operators don’t do this. Waymaker bakes it in so wins compound the way they did inside the org.

---

One thing for fellow operators

If you spent time in enterprise and you’re now building solo or small-team — most of what you learned still applies. The shape of the work has changed. The reasons it works haven’t.

The people who win the next decade are the ones who can apply enterprise discipline at solo speed. That’s not a contradiction. That’s a moat.

If you want to see where you stand on AI-readiness today, take the free 3-minute AI Skills Audit. It’ll route you to the path that fits where you are now.