IP Isn’t Your Moat.

Your Data and Workflows Are.

A few years ago, I was sitting across from a founder who had just come out of a very expensive patent battle. He had spent close to $400,000 in legal fees defending a core piece of IP that he’d built his entire go-to-market narrative around.

He won the case.

And six months later, a competitor built around it anyway — not by violating the patent, but by solving the same problem a different way. The moat he’d been defending didn’t exist. It was a line on a legal document. The market didn’t care.

That conversation stuck with me. Because I’ve watched a version of it play out over and over — founders and product leaders investing enormous resources into protecting something that feels like an advantage but isn’t actually one in the market.

AI just made that problem ten times worse. And most teams haven’t caught up to what that means.

The Old Playbook Was Written Before the Tools Changed

For the better part of two decades, the startup moat playbook read something like this: build something novel, patent it, protect it, and use that protection as a barrier to entry. Lock in customers through switching costs. Build your fundraising narrative around defensibility.

In fact, the angel investor syndicate I used to perform due diligence for prioritized this type of protection.

That playbook made sense when the cost to build was high, timelines were long, and technical complexity was itself a filter.

None of those conditions apply the way they used to.

Today, with the right AI tooling and a focused team, you can compress meaningful product development into weeks. I’ve seen it. I’ve done it. Competitors can too.

The timeline compression isn’t the scary part. The scary part is that everyone has access to the same compressor.

AI doesn’t just accelerate your development. It accelerates everyone’s. Which means the thing that used to buy you time — complexity — doesn’t buy you nearly as much anymore.

What AI Actually Collapsed

Let me be specific about what’s changed, because vague hand-waving about “AI disruption” doesn’t help anyone. It’s been overplayed.

Three things got dramatically cheaper and faster:

Building the first version. Foundational technical work that used to take a senior engineering team six months can now get to a functional prototype in a fraction of the time. Your competitors know this. AI-enabled product managers can build their own product’s MVP in weeks if not days.

Replicating a surface-level feature set. If your moat is “we have a better UI” or “we built this integration first” or “we have a proprietary algorithm” — assume someone can reverse-engineer the output, if not the method, inside of a product cycle.

Entering adjacent markets. The cost for an established player to spin up a competing offering in your category dropped substantially. This isn’t theoretical. I’ve watched it happen in fintech, in B2B SaaS, in lending infrastructure.

What AI didn’t collapse: the value of what you’ve already learned, what your customers have already told you, and how your team actually moves.

That’s the asymmetry most people are missing.

The New Moat Is What You’ve Accumulated, Not What You’ve Protected

Here’s the frame I keep coming back to: a moat is valuable because it’s hard to replicate quickly. The question isn’t “what can we protect legally?” It’s “what would take a competitor years to catch up to?”

The answer is increasingly: proprietary data and deeply embedded workflows.

Not the data you can buy. Not the publicly available training sets. The data that is a byproduct of how your customers use your product — the behavioral signals, the feedback loops, the edge cases that only show up at scale. The stuff that a competitor starting from zero simply doesn’t have, no matter how much they spend.

And workflows — the internal and external processes that your product is embedded in so deeply that switching isn’t just expensive, it’s disruptive to the actual operating rhythm of the customer’s business.

These are hard to replicate. They take time to build. They compound. And they don’t show up on a patent filing.

What This Looks Like in Practice

When I was working through a product launch into a market with significant regulatory complexity — financial services, where the surface area for compliance is enormous — the product itself was important, but it wasn’t the defensible asset. What was defensible was the decision logic we’d built from thousands of real scenarios, and the workflow integrations that meant our software was embedded in daily decisions across hundreds of institutions.

A competitor could build a competing product. They couldn’t instantly inherit decades of exception handling. They couldn’t recreate the muscle memory of thousands of loan officers who had been working in our system long enough that switching would mean retraining from scratch.

That’s a moat. And no patent attorney helped us build it.

Here’s how I think about it now — three questions every product and GTM leader should be asking:

1. Are we systematically capturing the data our product generates?

Not storing it — capturing it. There’s a difference. Because most of what I’m seeing is executives wanting to make data-driven decisions…but they don’t have the data. Storing is passive. Capturing means someone on your team is responsible for understanding what the data tells you about customer behavior, and feeding that back into the product and the GTM motion.

2. Is our product embedded in a workflow, or sitting adjacent to one?

Adjacent tools get cut. Embedded tools create switching costs. If your customer can turn off your product without changing how their team operates, you’re adjacent. If turning you off means rebuilding a process, you’re embedded. That’s the goal.

3. What does our data allow us to do that a new entrant simply can’t?

This is your actual positioning question. Not “what features do we have?” but “what can we learn, predict, or personalize, around our ICP, that is only possible because of the data flywheel we’ve been building?” If you can’t answer that with specificity, you don’t have a data moat — you have a data warehouse.

The Fundraising Implication Nobody Talks About

I’ve been in the room on a lot of capital raises — across $800M in debt and equity. I’ve watched how investor conversations have shifted.

Three years ago, a strong IP story could carry a lot of weight in a deck. “We have 12 patents pending” read as defensibility. Smart investors always knew it was mostly signaling, but it played.

Today, the investors I respect are asking different questions. They want to know what data you have that nobody else has. They want to understand how embedded you are in your customer’s workflow. They want to see evidence that your product gets more valuable as you accumulate more customers — not just bigger, but smarter.

That’s a data flywheel story. That’s a workflow entrenchment story.

If your defensibility narrative is still built around IP, you’re bringing a 2018 pitch to a 2026 room. That’s not a minor update to your deck — it’s a strategic rethink.

The Uncomfortable Honest Take

Here’s the part that’s hard to say, but important: most early-stage companies haven’t built a data moat yet. They have data. That’s different.

Accumulated data without a system for learning from it isn’t a moat. It’s a liability — storage costs, compliance exposure, and a false sense of security.

The companies building real moats right now are the ones who are intentional about it from day one. They’re thinking about what behavioral data gets captured at every touchpoint. They’re building for workflow integration, not just feature differentiation. They’re treating their data infrastructure as a strategic asset, not an engineering afterthought.

That’s a founder mindset decision. It doesn’t require a massive team or a massive budget. It requires choosing to treat data as a product.

Where to Start This Week

If you’re a founder or product leader reading this, here’s what I’d actually do:

Audit your data touchpoints. List every interaction your customer has with your product. Now ask: which of these generate data we’re actively learning from? Which are we just logging and forgetting?

Map your workflow depth. For each customer segment, draw the workflow your product sits inside. Be honest about whether you’re embedded or adjacent. If you’re adjacent, what would it take to get embedded?

Find one data advantage to name. If you can’t articulate a specific thing you know — about your customers, your market, or your product’s performance — that a competitor couldn’t learn in six months, you don’t have a data moat yet. That’s your starting point.

The window to build this isn’t infinite. Your competitors are figuring this out too.

The good news: data and workflow advantages compound over time. The earlier you start being intentional about them, the harder you become to catch.

IP will still matter in some contexts. But it’s no longer the thing that protects you in the market. What protects you is what you’ve learned and how deeply you’re woven into how your customers operate.

Start there.