What If the Product Is Not the App

There’s a question that keeps coming up in how we talk about product and AI, and it feels bigger than tooling or features: should we still be building SaaS applications, or should we be building AI skills and agents that people license?

Not chatbots bolted onto an app. Not “AI-powered” labels. Actual capabilities. Intelligence that can live anywhere. GPTs, Claude skills, or agents that people pay for because they carry real domain knowledge and automation power.

It echoes what happened years ago with platforms like Salesforce. First came full applications. Then the platform stabilized. And the value shifted. People stopped rebuilding entire systems and started building modules, plugins, and workflows inside an ecosystem. The application became the foundation. The differentiation moved to what you could build on top of it.

AI feels like the beginning of a similar shift.

I read that Walmart gave all their corporate employees access to ChatGPT and are heavily investing in AI fluency across their workforce. If that’s happening at scale, it feels natural to assume that knowledge work and office work will start shifting to text-based interfaces like ChatGPT and Claude. The conversation becomes the workspace.

Most product thinking still starts with the application: a UI, a workflow, a database, a subscription. But AI introduces something different. Capabilities that don’t need to live in one interface. Logic that can be called from anywhere. Intelligence that feels modular and portable.

So the question becomes less about what app is being built, and more about what intelligence is being created.

Instead of paying for access to software, people may start paying for access to reasoning, automation, and domain expertise. Not “use our product,” but “use how we think.”

That starts to look like a capability economy, not just a SaaS economy.

SaaS Still Matters

SaaS still matters. It still owns the things that are hard and essential: trusted data, permissions, workflows, governance, auditability, and long-term memory. AI agents don’t replace that. They depend on it.

Without a platform:

• Where does the data live?
• Who controls access?
• Who ensures correctness?
• Who maintains history?

So it isn’t SaaS versus AI agents. It’s SaaS becoming the substrate for intelligence.

The platform becomes where data is trusted and rules are enforced. AI becomes where reasoning happens, insight is generated, and work is accelerated.

The Questions Change

Which means when platforms are built today, they aren’t just products. They are environments where intelligence can exist and evolve.

That changes the questions:

• How does AI access trusted data?
• How is it governed?
• How is it licensed?
• Can it exist independently of the UI?
• Can someone buy the intelligence without buying the application?

That last one is uncomfortable and exciting.

Could someone license what you know without touching your interface?

That’s not a feature decision. That’s a business model decision. That’s an identity decision.

From Software to Capability

SaaS sold software. What feels like it’s emerging now is selling capability.

And just like strong platforms became more valuable when ecosystems grew around them, platforms that make space for intelligence will become more valuable, not less.

The Ground Is Shifting

The ground is shifting beneath us. Will it stay that way, or will we be constantly chasing new paradigms?

Maybe we’ll be building modules that are only visible to you if you have permissions to see the data, appearing as you walk around. Intelligence layered onto the world, governed by trust and access.

Fun to imagine, isn’t it?