When Building Gets Easy, What's Left to Defend?

The first thing you realise after building using AI… is that any of the hundreds of millions of people using AI can build the exact same thing.

The quality of AI coding tools has improved conceptually, to the point where it can really no longer be considered 'Vibe Coding"; implying that the human does all the planning and the agent just outputs text. We now have agents that can plan, architect, debug, design, and interact with complex codebases.¹ It is now better described as "AI Augmented Development" or "Vibe Development".

Using AI tools is no longer limited to wannabe's such as myself who share their latest project via LinkedIn at http://localhost:3000, it is being used effectively by the best coders in the world on real projects.²

It got me thinking, that as Vibe-developing capability increases, what moats exist for new and established software products?

Traditionally, companies would be able to build out proprietary codebases. The features this codebase creates add product value to customers, Company gets more customers who pay more money.

This used to be a defensible business model, for a number of reasons:

Note: This list is not exhaustive.

1. Effort to recreate - Accumulated Complexity

Creating code was bottlenecked by human processing. A sophisticated feature might represent thousands of hours of engineering time. Even if a competitor knew what you'd built, recreating it meant hiring expensive engineers and waiting months or years.

In a world where AI capability rapidly increases development output, products can accumulate complexity and features at a pace that might soon be able to keep up with an incumbent company, with less overheads.

2. Talent Scarcity

The people who could build these systems were rare and expensive. Companies that attracted top engineering talent could execute on ideas that others simply couldn't staff for.

AI augmented development will not only increase the capability of existing developers, and reduce the bottleneck of human processing, but it actually removes barriers for people to enter the market, in the same way that drag and drop software opened up doors for many. (More on this in future articles)

3. Intellectual Property Law

Patents, trade secrets, and copyright created legal barriers around implementation details. Even if someone could theoretically rebuild your system, they risked litigation. This gave companies time to establish market position while competitors either licensed the technology or worked around it.

AI does not remove this barrier, but ideas are harder to patent, and the software industry has always been an interesting legal area for IP. For example, what code is proprietary, and what is simply the obvious way to implement a well-understood function? The Oracle v. Google case spent a decade litigating whether APIs themselves could be copyrighted. The answer was murky at best.

AI makes this murkier. When AI generates code, it's not copying your implementation, it's producing functionally equivalent output from a statistical understanding of how software works. You can't sue someone for independently arriving at the same solution, and that's essentially what AI-assisted development enables at scale.

The moat was never really the code. It was the institutional knowledge required to write it. That's what's evaporating.

4. Product Hegemony - Usability through network effect

Microsoft for example built it's early hegemony in software by becoming the no-brainer platform for early micro-processor and PC development, becoming the platform that everybody used, and knew how to use.³

AI presents the possibility of a user interface which can be fully customisable in terms of its capability, and usability. Why would any business or user turn to horizontal or generic software, when they could have a fully customised version, down to a user level?⁴

(Note, This is in reference to the Microsoft business model in the very early days, not today)

All of this accumulates to put pressure on SaaS providers as the sellers. With the price of software development down, It now affords businesses and consumers options to:

A. Replace subscriptions with a cheaper new entrant

B. Build the software or elements of it themselves⁵

C. Use the leverage created by those options to negotiate a better contract.

The same equation exists in the M&A space in company valuation.

For enterprises who place a premium on reliability, security, or have complex existing integrations, this equation might not make sense yet. But looking ahead to the future, I can only see the capability of vibe-developing increasing.

None of this means software companies disappear overnight. The strong ones built moats around things other than code—data, distribution, trust, relationships. (I'll dig into these in future articles.) What's changing is how much time code complexity alone actually buys you.

People inside the industry probably know this already. But for the businesses and consumers still operating like software is scarce: The power dynamic has shifted, and it's shifting in your favour.

When Building Gets Easy, What's Left to Defend?

1. Effort to recreate - Accumulated Complexity

2. Talent Scarcity

3. Intellectual Property Law

4. Product Hegemony - Usability through network effect

Sources:

More interesting reading: