Clayton Davis Blog

The blog of an overly opinionated cloud consultant

16 Nov 2026

Anyone Can Build an App. That's Why Product and Design Matter More Than Ever.

AI is fundamentally changing the economics of software development. What used to take months can now be prototyped in days. What used to require a team of engineers can sometimes be accomplished by one person with Claude and a good prompt.

That’s incredible. But it also creates a problem.

When everyone can build software, simply building software is no longer a competitive advantage.

I believe this is going to fundamentally change how we think about application development. As AI makes engineering faster and more accessible, the bottleneck shifts. It’s no longer just about how quickly we can turn requirements into code. It’s about knowing what to build, who we’re building it for, and how to create an experience people actually want to use.

In other words, product thinking and design are becoming more important in an AI-powered software development lifecycle, not less.

And before you assume I’m advocating for spending three months on product discovery before anyone writes code, I’m not. Quite the opposite. I think AI gives us an opportunity to rethink how product, design, and engineering work together.

But first, let’s look at what’s happening in the market.

The App Store Has a Problem. And It’s Coming for Enterprise Software.

The number of applications being created is exploding, but the amount of attention people give those applications isn’t growing at nearly the same rate.

I was listening to Dylan Field, CEO of Figma, on a recent episode of the All-In podcast, and he brought up a statistic that really stuck with me.

The data backs him up.

According to Sensor Tower data reported by The New York Times, in the first half of 2026:

  • New applications released on Apple’s App Store increased 100% compared to the same period in 2025.
  • App downloads increased just 2%.

Let that sink in. Twice as many new applications entering the market, with almost no corresponding growth in downloads.

And this isn’t limited to the broader App Store. RevenueCat’s 2026 State of Subscription Apps report found that monthly new subscription app launches grew from approximately 2,000 in January 2022 to more than 14,700 in January 2026. That’s more than a 7x increase.

Yet applications launched before 2020 still account for 69% of subscription revenue in RevenueCat’s dataset. Apps launched in 2025 or later account for just 3%.

Now, to be fair, older applications have had years to build an audience. And downloads aren’t the same thing as usage. Sensor Tower’s broader mobile research actually shows time spent in applications growing 3.8% in 2025.

But that’s kind of the point.

The supply of software is growing exponentially faster than the attention available to consume it.

AI has removed much of the friction from creating software. What it hasn’t done is give people more hours in their day or more reasons to care about another application.

We’ve spent years asking how to build software faster. I think we need to start asking a different question.

Why would someone want to use what we’re building?

The Rise of the Perfectly Average Application

Let’s talk about Claude for a minute.

I’m a huge believer in AI-powered engineering. I’ve written previously about how AI is changing the software development lifecycle, and I think we’re still in the early stages of understanding what’s possible.

Today, I can describe an application, give Claude some requirements, and have a working interface in a fraction of the time it would have taken me a few years ago.

That’s incredible.

But there’s a catch.

Everyone else can do the same thing.

Give Claude a generic prompt to build a modern SaaS application and you’re likely to get something that looks pretty good. Clean typography, a sidebar, some nicely spaced cards, a few charts, maybe a purple gradient for good measure.

It’s functional. It’s polished. It might even be impressive.

It’s also probably not very different from the next hundred applications being generated with similar prompts.

Interestingly, Anthropic has acknowledged this themselves. In their March 2026 engineering article on harness design for application development, they observed:

“Absent any intervention, Claude normally gravitates toward safe, predictable layouts that are technically functional but visually unremarkable.”

That’s a pretty important statement coming from the people building the tool.

And it’s not just Anthropic. Microsoft Research has explored how AI-powered website generation risks reinforcing common design patterns rather than encouraging originality. Research published in Science Advances similarly found that AI assistance improved the quality of individual creative work, but also made the outputs more similar to one another.

To be clear, these studies don’t prove that every AI-generated interface looks alike. But they highlight a real risk.

When everyone has access to the same powerful tools, accepting their default output is unlikely to create a competitive advantage.

And I don’t think the answer is to stop using Claude for design. Quite the opposite.

The problem isn’t that AI can’t create great designs. It’s that great design requires direction, context, and a point of view.

Claude can help execute that vision. It can generate alternatives, prototype interactions, explore layouts, and accelerate iteration.

But someone still needs to decide what makes the experience special.

The purple gradient isn’t the problem. The absence of an opinion and expertise is.

Product Tells You What to Build. Design Makes It Worth Using.

When people hear “design,” I think they often jump straight to visual design. Colors, fonts, buttons, and making things look pretty.

That’s certainly part of it, but it’s a pretty small part of what makes a great product.

Let’s separate two disciplines that I think become increasingly important as engineering accelerates.

Product: Are We Building the Right Thing?

Product thinking starts well before anyone writes a prompt or a line of code.

Who are we building for? What problem are they trying to solve? How are they solving it today? What makes our approach meaningfully better?

These sound like obvious questions, but how often do we actually answer them before engineering begins in the era of AI powered engineering?

I’ve seen plenty of projects start with a list of features rather than a clearly defined problem. The assumption is that if we build the requested functionality, we’ve delivered value.

Sometimes that’s true.

Other times, we’ve just built exactly what someone asked for, without understanding what they actually needed.

AI makes this problem worse, not because AI creates bad software, but because it lets us create the wrong software incredibly efficiently.

Think about it. If I can build an application in a week instead of three months, that’s a massive win.

Unless nobody wants to use it.

Now I’ve just wasted a week instead of three months. Better, certainly, but still not the outcome we were hoping for.

Product strategy gives engineering something meaningful to aim at. It defines the user, the problem, the expected outcome, and how we’re going to measure success.

Design: Does the Experience Actually Deliver?

Once we’ve established what we’re building and why, design becomes the bridge between that strategy and the experience people interact with.

And I’m not just talking about consumer applications.

Imagine an internal enterprise application that takes a 12-step workflow and turns it into three steps. Or a customer-facing platform that makes a complex process intuitive enough that someone can complete it without training.

That’s great design.

It understands how people think, what they’re trying to accomplish, and what gets in their way.

Great design also creates differentiation. It gives a product a personality, builds trust, and makes people want to come back.

And it has measurable business implications.

In its Business Value of Design research, McKinsey found that companies scoring in the top quartile for design practices experienced 32 percentage points more revenue growth and 56 percentage points more shareholder-return growth than industry benchmarks over five years.

That study is from 2018, and it’s correlation rather than proof that design caused those outcomes. But the underlying principles remain relevant: user understanding, cross-functional collaboration, measurement, and continuous iteration.

Those aren’t aesthetics. They’re product development capabilities.

AI Doesn’t Eliminate Product Discovery. It Changes How We Do It.

Here’s where I want to be careful, because I can already hear the objection.

“Great, so now that engineering takes two weeks, we’re going to spend three months doing product discovery and design?”

Absolutely not.

I’ve written before about the shift from traditional strategy-led development to a hybrid approach. AI fundamentally changes the economics of iteration.

When building a prototype was expensive, it made sense to spend significant time planning before engineering started.

Today, we can put functional software in front of users incredibly quickly.

That should change how product, design, and engineering collaborate.

Instead of lengthy sequential phases, I think the winning model is a tight, continuous loop:

  1. Understand the problem. Identify the users, their needs, and the business outcome we’re trying to achieve.
  2. Form a product hypothesis. Define what we believe will solve the problem and why.
  3. Design and build quickly. Use AI-powered engineering to turn that hypothesis into something tangible.
  4. Validate with real users. Find out what works, what doesn’t, and what we misunderstood.
  5. Refine and repeat. Feed the learning back into product, design, and engineering.

Think days and weeks, not quarters.

And yes, AI should be part of all five steps. It can help synthesize research, explore potential solutions, generate design alternatives, and build working prototypes.

But there’s a distinction between using AI to accelerate product discovery and using AI to replace it.

A recent Nielsen Norman Group study compared responses from 604 human participants with 120 AI-generated respondents. While both groups reached broadly similar conclusions, humans produced more than 50 distinct response labels compared to just 12 from the synthetic participants.

It was a narrow research exercise with methodological limitations, but the finding illustrates something important.

AI can be very good at telling us what a typical user might say. The unusual answers, unexpected frustrations, and edge cases are often where the interesting product opportunities live.

Those are the insights we need to find, not smooth over.

The Bottleneck Is Moving Left

For decades, engineering capacity was one of the biggest constraints on software delivery.

We had more ideas than we could implement, so organizations optimized around getting features through engineering as efficiently as possible.

AI is changing that equation.

As engineering throughput increases, the bottleneck moves upstream.

More specifically, it moves toward the people defining the problems, making product decisions, and designing the experience.

If engineering can build ten features in the time it previously took to build two, someone needs to make sure those ten features are worth building.

Otherwise, we’ve just created a faster feature factory.

This is why I believe product management and design aren’t becoming less important in an AI-powered SDLC. They’re becoming more important.

But they also need to evolve.

Product managers need to become better at turning customer understanding into clear, AI-consumable context. Designers need to use AI to explore and validate ideas faster, rather than simply producing static handoff artifacts. Engineers need to be involved earlier in shaping solutions, not just implementing decisions someone else already made.

The goal isn’t more process.

It’s better decisions, made earlier, with faster feedback.

Building Software Is No Longer the Finish Line

I’ll leave you with one more thought from Dylan Field.

During that All-In interview, he described design as “the new code.”

I don’t take that to mean engineering is unimportant. Secure, scalable, reliable software is still hard to build. AI doesn’t magically eliminate architecture, integration complexity, or operational responsibility.

But I do think he’s right about where differentiation is moving.

For years, simply being able to build sophisticated software was a competitive advantage.

Now that capability is becoming increasingly accessible.

The next advantage comes from combining three things:

  • Product judgment: Knowing what problems are worth solving.
  • Intentional design: Creating experiences that are differentiated, useful, and memorable.
  • AI-powered engineering: Turning those decisions into reliable software faster than ever before.

That’s the combination I’m excited about.

And it’s exactly the philosophy behind what we’re building at Caylent.

Our Cloud Native App Dev offering brings product discovery, experience design, and AI-accelerated engineering together to help organizations move from an idea to a validated, production-ready application.

Not just to ship more code. To build something people actually want to use.

Because in a world where anyone can build an app, the real competitive advantage is building one worth using.