The Incredible Shrinking Product

In the age of AI, the lines between a single feature and a complete solution are blurring faster than ever.

Remember Microsoft Word in the 90s? It was a revolutionary product. Today, its core text formatting capabilities are a standard feature bar in countless apps. What was once a destination is now just a step on a longer journey. To its credit, Word kept adding capabilities, evolving over decades into the robust suite it is today. But that evolution took years. In the age of AI, we don’t have that luxury - product evolution is happening much faster.

The conversation around building with AI is intense. I’ve often said that we need to think bigger, to lean into second-order thinking and set ambitious North Stars. And while that’s crucial, in our rush to the future, we’re at risk of forgetting a fundamental rule of building: you still have to make something people actually want.

AI is not the product. It’s the engine. It’s a capability that can help you solve a user’s problem in a way that was never before possible. But if you’re not relentlessly focused on the user problem itself, all you have is a cool tech demo.

From Capability to Solution

With models becoming more powerful by the day, the temptation is to simply expose their capabilities and call it a day. But a thin wrapper around an API isn’t a product; it’s a commodity waiting to happen. To build something that lasts, you need an opinionated take on the value you’re providing.

Consider image generation. On its own, it’s not a product.

It’s an incredible, mind-bending capability. When they first appeared, standalone image generators felt like magic. They were revolutionary, and for a moment, they were the product. But technology that starts as magic quickly becomes an expectation. Today, that novelty has faded. Now, a standalone image generator can feel incomplete - like a toy - while an image generation feature inside Google Slides or Canva feels like a superpower.

The capability is the same, but the context is everything. The feature is seamlessly integrated into the user’s existing workflow, meeting them exactly where they are.

Mapping the Entire Journey

The best way to escape the trap of building for a single moment in time is to map the user’s entire journey. To do that, ask yourself two simple questions:

• What was my user doing right before this?

• What do they need to do right after?

Then, keep asking. Follow the chain of tasks forward and backward until you have a clear picture of the entire workflow, not just the single step you’re focused on.

Think about the journey of Adobe Firefly. It launched as a standalone tool to access their image model, and it was a hit. But the question of “what comes after” became critical. Okay, I have an image. Now what? Do I need to turn it into a poster, use it in a newsletter, or create an ad? This is where a product like Canva shines. It understands the “after.”

But just as important is the “before.” Before a user needs an image for a social media campaign, they were probably brainstorming copy or defining their target audience. By understanding the full context, you move from solving a single step to owning the entire workflow.

The Great Workflow Consolidation

This focus on the complete journey is causing the lines between products to blur. Workflows that once required hopping between five different tabs are now being consolidated into a single, seamless experience.

We’re seeing this play out in real-time. Conversational AI tools like Perplexity are no longer just for research. They’re integrating purchase flows directly into the experience. The user journey - from discovering a need, to researching solutions, to making a purchase - used to be fragmented across multiple platforms. Now, it can happen in one place.

The end-to-end journey is becoming the product.

Where Do We Go From Here?

As builders, this leaves us with two exciting paths forward.

1. Find a broken workflow and fix it. Look for user journeys that are clunky, inefficient, and require too many handoffs between different tools. What steps can you unify? Your wedge might be solving one part of that journey 10x better than anyone else, then expanding outward until you’ve streamlined the whole process.

2. Discover a brand-new workflow. AI doesn’t just improve old processes; it makes entirely new ones possible. What new user needs and behaviors will emerge as these tools become more integrated into our lives? Building for these novel journeys is how you define the next category instead of just competing in the current one.

Ultimately, it all comes back to a simple truth. Study the models. Develop an intuition for what they can do. But then, turn away from the technology and look squarely at the user.

Build something that solves a real, human problem in a way that feels delightful. Because the most successful products won’t be the ones with the flashiest tech. They’ll be the ones people can’t imagine their lives without.