Selling in the Age of AI: If Everyone Can Build Software, What Are We Actually Selling?

“I could just build that myself.”

It's something I suspect we're going to hear increasingly often as artificial intelligence changes the economics of software development.

And in many cases, it's true.

With tools such as Claude and other AI coding assistants, people can describe an idea, generate code, build an application and create something that would previously have required a developer or an entire development team.

The barriers to building software are falling rapidly.

But I think this creates an interesting question for anyone developing technology products:

If customers can build the software themselves, what exactly are they paying us for?

I believe we're looking at a fundamental shift in how technology creates and captures value.

1. Building software is not the same as running a business

Let's take a small business owner who wants an AI assistant to manage customer enquiries, organise appointments, respond to questions and help with day-to-day administration.

They could open Claude, describe what they want and ask it to build an application.

They might even produce something that works.

But building an application is only one part of the problem.

What happens when the application needs to connect to other systems? Who manages the agents? How are permissions assigned? What information can each agent access? How are errors handled? Who maintains the underlying infrastructure? What happens when something breaks or the business needs support?

And perhaps most importantly, how does the technology understand the context in which the business operates?

There is a considerable difference between having a piece of software and having a functioning business environment.

2. The product is becoming the ecosystem

I think the value proposition of technology platforms is changing.

Historically, customers paid for software functionality. They purchased a product that performed a particular task: accounting, customer management, booking, marketing or communication.

Increasingly, AI allows these capabilities to be created, adapted and connected much more easily.

This could make individual software features less differentiated.

The value may therefore shift towards the environment in which those capabilities operate.

Consider what a business is actually paying for when it subscribes to an AI-enabled business platform:

• The agents: The intelligent assistants designed to perform particular business activities.
• The configuration: The setup, workflows and instructions that make those agents useful.
• The infrastructure: The backend systems, integrations and technical environment that allow everything to operate.
• The governance: The rules, permissions, access controls and safeguards governing what agents can and cannot do.
• The orchestration: The coordination between different agents, tools, systems and business processes.
• The maintenance and support: The ongoing management, monitoring, improvements and troubleshooting that the business owner does not need to handle themselves.

Much of this work is invisible to the customer.

And that is precisely the point.

The customer isn't necessarily paying for something they couldn't build. They're paying to avoid having to build, manage and maintain it themselves.

3. The value of collective intelligence

There is another dimension that I think could become particularly important: knowledge accumulated through collective activity.

Imagine an AI agent designed to help a small business manage customer enquiries.

One business owner can build an agent and teach it how to respond to their customers. But that agent initially has a limited context. Its knowledge comes from the instructions, information and experiences available to that individual business.

Now imagine an ecosystem where many similar businesses use agents operating within a shared platform.

Over time, the platform can learn from patterns across different businesses.

Which types of enquiries are most common? Which questions repeatedly cause confusion? Which workflows tend to be effective? Where do businesses encounter friction? Which processes could be improved?

Subject to appropriate privacy protections, data governance and controls, aggregated patterns and shared improvements can inform how the ecosystem develops.

An individual business owner may be able to build a similar agent, but they cannot easily replicate the accumulated learning of an entire community of businesses working within the same environment.

This is a different kind of value.

It is not simply artificial intelligence. It is intelligence that improves through collective experience.

There is an important distinction here: this does not mean businesses should have access to one another's confidential information. Shared intelligence needs to be designed carefully, with appropriate separation of customer data, permissions and privacy safeguards.

The opportunity is to learn from collective patterns without compromising individual businesses.

4. The invisible infrastructure is part of what customers buy

We often talk about AI as though it exists independently.

We ask an AI model a question, receive an answer and see the intelligence in front of us.

But when AI starts operating within a business, there is an entire infrastructure behind that interaction.

Agents need identities. They need access to information. They need defined responsibilities. They need to know which systems they can interact with and which actions require approval.

They need to operate within boundaries.

There must also be a way to monitor activity, manage failures, update configurations and maintain the underlying systems.

This becomes particularly important as businesses move from using individual AI assistants to deploying multiple agents that interact with one another and perform connected activities.

The challenge is no longer simply creating intelligent agents. It is managing an ecosystem of agents safely and reliably.

That is a substantial undertaking, and one that many small businesses will have neither the time nor the expertise to manage independently.

5. From selling tools to selling outcomes

I suspect this will change how technology companies position and price their products.

Customers may become less interested in paying for individual features and more interested in paying for an outcome.

A small business owner does not necessarily want to purchase five different AI agents, configure them, connect them to their systems and manage their permissions.

They want their business to operate more efficiently.

They want customers to be able to find them, enquiries to be handled, appointments to be organised and routine administration to become easier.

The technology should make that happen without requiring the business owner to become a technology specialist.

This is where the subscription model becomes interesting.

A subscription is not simply a payment for access to software. It can represent access to an evolving ecosystem: the agents, infrastructure, configuration, governance, collective learning and ongoing support that make the service work.

The platform takes responsibility for managing the complexity so that the customer can focus on running their business.

6. The new competitive advantage

If AI continues to make software development more accessible, the ability to write code may become a less significant competitive advantage on its own.

That doesn't mean software engineering becomes unimportant. Building reliable, secure and scalable systems will remain difficult.

But I think the differentiators will increasingly include:

• The quality of the ecosystem being built.
• How effectively different agents and systems work together.
• The governance and permissions framework.
• The knowledge accumulated through usage and experience.
• The quality of the business context in which the technology operates.
• The ability to maintain, support and continuously improve the environment.

A competitor may be able to reproduce an individual feature. Reproducing the entire ecosystem, including its operational infrastructure, accumulated knowledge, relationships and ongoing improvements, is a much harder proposition.

The competitive advantage moves from owning a piece of software to managing an environment that becomes more valuable as it evolves.

7. So, what are we really selling?

I keep coming back to the original statement:

"I could just build that myself."

Yes, you probably could.

But would you want to spend your time building it, connecting it, configuring it, governing it, maintaining it and continuously improving it?

And could you independently reproduce the knowledge accumulated through thousands of interactions across a community of similar businesses?

There will always be people who prefer to build their own solutions. AI will make that increasingly possible, and that is a positive development.

But there will also be people who would rather pay for a system that works, evolves and takes care of the complexity on their behalf.

I think this is where the next generation of technology businesses will find their value.

We may be moving from an economy where companies sell software to one where they sell managed intelligence, connected environments and the collective knowledge that develops within them.

In the age of AI, perhaps the most important question is no longer whether someone can build the software themselves.

It is whether they should have to.

What do you think? As AI makes software increasingly accessible, will the real value shift from the applications themselves to the ecosystems that manage them?