We learned nothing from SaaS lock-in.

We watched as “pay for what you use” became “pay forever or lose your data.” We saw “focus on your business” turn into “focus on our quarterly earnings.” We experienced “best practices” degrade into “best for our ecosystem.”

Now we’re doing it again with AI. Faster this time.

The Wrong Fear

Every organization I talk to is worried about AI seeing their data.

They’re nervous about training on confidential documents. Concerned about privacy. Anxious about what happens to their information once it enters an AI system.

These are legitimate concerns. But they’re not the biggest threat.

The biggest threat is losing control of your organization’s ability to think.

Not to AI itself. To the vendors providing it.

The Shopifyification of Intelligence

Remember when Shopify was going to democratize e-commerce? Let anyone compete with Amazon?

It did. Sort of.

Now every Shopify store looks basically the same. Same templates. Same checkout flow. Same features. Same limitations.

You can succeed on Shopify. Plenty of businesses do. But you’re not differentiated by your storefront anymore. You’re differentiated by your marketing, your products, or your brand—not by anything about how your actual store works.

Shopify optimized for adoption. Make it easy for everyone. Which means advantages for no one.

AI is about to do this to your intelligence infrastructure.

Microsoft Copilot, Google Gemini, Salesforce Einstein—they’re all offering the same promise: integrate with us, we’ll make your organization smarter, everyone’s doing it.

Five years from now, every company in your industry will be using the same AI reasoning over the same templated understanding of what a business should look like.

Your AI will give you the same insights as your competitors’ AI. Because you’re both using Microsoft’s generic model of business operations.

Congratulations. You’ve automated your way to strategic mediocrity.

Three Threats Nobody’s Talking About

While everyone worries about data privacy, three bigger problems are being ignored:

1. Vendor Lock-In at the Intelligence Layer

When your “AI strategy” is really just “Microsoft Copilot” or “Google Gemini,” you’re not building AI capability. You’re renting it.

Your organization’s knowledge gets encoded in their system. Your workflows get built around their tools. Your team gets trained on their interface.

Two years later, they raise prices. Or deprecate features. Or decide your industry isn’t strategic. Or get acquired by someone who changes everything.

And you realize: switching would cost more than just staying.

You don’t have an AI dependency. You have an intelligence dependency.

Your organization’s ability to reason about its own operations is now tied to their roadmap, their pricing decisions, and their definition of what your data means.

Security breaches can be contained and fixed. Vendor lock-in is designed to be permanent.

2. Competitive Commoditization

This is the Shopify problem.

When everyone uses the same substrate—the same templates, the same “best practices,” the same generic understanding of business—AI stops being a competitive advantage.

Your competitor gets the same insights you do. Makes similar recommendations. Spots the same opportunities. Identifies the same risks.

Because you’re both reasoning over Microsoft’s idea of how businesses work, not your unique understanding of how your business works.

The vendor optimizes for broad adoption. Make it work for everyone in every industry. Solve the common problems.

Which means it advantages no one.

This is the “best practices” trap all over again. Everyone achieves parity by following the same templated approach. Nobody gets differentiated. Everyone descends into bland, competent mediocrity.

The companies that win with AI won’t be the ones using the best models. They’ll be the ones whose AI reasons over their unique understanding of their domain.

3. The Enshittification of Intelligence

We’ve seen this pattern with every platform:

Stage 1: Be great to users (gain adoption) Stage 2: Be great to business customers (monetize) Stage 3: Be great to shareholders (extract maximum value)

Social media platforms did this. SaaS tools did this. Cloud providers are doing this. Search is doing this.

AI will follow the same arc.

Right now, AI vendors are in Stage 1. Making it amazing. Easy to adopt. Generous terms. Impressive capabilities.

Stage 2 is already starting: enterprise tiers, business features, integration ecosystems, vendor partnerships.

Stage 3 is inevitable: price increases, feature deprecation, forced migrations, reduced support, optimization for revenue over user value.

Except this time, it’s not just your tools that degrade. It’s your organization’s ability to think.

When your intelligence infrastructure enshittifies, you can’t just complain on Twitter and switch to a competitor. Because your organizational knowledge is encoded in their system.

What You’re Actually Losing

This isn’t about the technology. It’s about control.

You’re losing control of your substrate.

Your substrate is the organized, contextualized representation of your organization’s knowledge. It’s not just your data—it’s the meaning of your data. The relationships. The context. The “why” behind the “what.”

Every AI needs a substrate to be useful. The model is just the reasoning engine. Your substrate is what it reasons over.

Here’s the trap: The vendors want to build your substrate for you. Inside their platforms. Using their tools. Locked to their APIs.

Once they do:

• You can’t easily switch to a better model
• You can’t customize how AI understands your domain
• You can’t encode your proprietary advantages
• You can’t port your intelligence to new systems
• You can’t escape without starting over

You become dependent on their definition of what your organization knows.

The Intelligence You Can’t Reclaim

The scariest part isn’t the data you give them. It’s the understanding you build in their system.

Over time, your organization learns to work with their AI:

• Your workflows adapt to their capabilities
• Your team learns to phrase things their way
• Your processes assume their features
• Your strategy depends on their insights

This isn’t just vendor lock-in. This is cognitive lock-in.

Your organization’s intelligence—how it thinks, how it reasons, how it makes decisions—becomes shaped by their system.

Five years from now, someone asks: “Why do we do it this way?”

The answer: “Because that’s how Copilot works.”

Not because it’s the best way. Not because it’s your competitive advantage. Because it’s what the vendor’s system made easy.

You’ve outsourced not just the execution, but the thinking.

The Shopify Store Problem

Shopify stores can be successful. Many are.

But they succeed despite the platform, not because of it.

Their differentiation comes from marketing, brand, product—not from anything about how the store actually works. The store is just competent and generic.

AI will do this to your entire operation.

Your workflows will be competent and generic. Your insights will be competent and generic. Your decision-making will be competent and generic.

Just like every other company using the same vendor’s intelligence infrastructure.

The question isn’t whether you’ll survive. The question is whether you’ll thrive.

Plenty of businesses survive on Shopify. Fewer truly thrive because of it.

What Independence Actually Means

AI independence means your organization’s ability to reason isn’t tied to any single vendor.

It means:

• You can switch AI models when better ones emerge
• You can use different models for different purposes
• You can encode your unique competitive advantages
• You can customize how AI understands your domain
• You can leave without losing your intelligence

It means your substrate—your organizational knowledge and context—belongs to you.

Not Microsoft. Not Google. Not OpenAI. You.

The AI model is just the engine. You should be able to swap engines without rebuilding the whole car.

The Window Is Closing

Right now, the AI ecosystems are relatively open. You can still build independence.

In two years, it will be harder. The platforms will be more closed. The switching costs will be higher. The network effects will be stronger.

In five years, you’ll be looking at consultant bills and eighteen-month migration projects you can’t afford.

The companies that thrive won’t be the ones who picked the “right” AI model early.

They’ll be the ones who kept their independence while everyone else was chasing convenience.

The Real Question

Everyone’s asking: “Which AI should we use?”

That’s the wrong question.

The right questions are:

“Who controls our organization’s intelligence?”

“Can we leave without losing our ability to think?”

“Are we building capability or renting it?”

“Will our AI give us unique advantages or generic competence?”

Security threats can be managed. Privacy concerns can be addressed.

But vendor lock-in at the intelligence layer? That’s giving up control of how your organization reasons about itself.

That’s not an IT decision. That’s a strategic surrender.

What We’re Declaring

We declare that:

Organizations should control their own intelligence infrastructure.

Not rent it from vendors who can change the terms.

AI should amplify competitive advantages, not eliminate them.

Not reduce everyone to the same templated mediocrity.

Models should be replaceable.

Not locked to the substrate they reason over.

Independence should be designed in from the start.

Not attempted as a rescue mission five years too late.

This Isn’t Anti-AI

This isn’t a rejection of AI. AI is transformative. The capabilities are real. The opportunities are enormous.

This is about recognizing that we’re making the same mistake we made with SaaS.

We’re trading long-term control for short-term convenience.

We’re accepting vendor lock-in because integration is easy.

We’re adopting “best practices” that guarantee mediocrity.

We’re building dependencies we won’t be able to escape.

Except this time, we’re not just outsourcing our infrastructure.

We’re outsourcing our intelligence.

The Choice

You can build on their substrate. It’s easier. Faster. More convenient.

You’ll get AI capability quickly. Your team will be productive. You’ll see results.

And in five years, you’ll be:

• Paying whatever they charge
• Using whatever features they offer
• Competing with whatever capabilities they provide to everyone
• Reasoning the way they designed you to reason

Or you can build your own substrate. It’s harder. Slower. More expensive upfront.

But in five years, you’ll be:

• In control of your intelligence
• Able to use any AI model that serves your needs
• Encoding competitive advantages your competitors can’t copy
• Reasoning in ways that reflect your unique understanding

The companies that keep their independence will have options. The rest will have dependencies they can’t escape.

Start Now

The time to think about AI independence is before you’re dependent.

Not after your organization has spent two years building on someone else’s substrate.

Not after your team has trained on their workflows.

Not after your strategy depends on their insights.

Not after switching would cost more than staying.

Before the vendors build your cage, build your foundation.

The smart companies are already thinking about this.

The rest are about to learn the same lesson they learned with SaaS:

Freedom is something you build, not something you subscribe to.

---

This is not a product. This is not a service. This is a warning about what we’re about to lose if we’re not careful.