The ecosystem is the disruptor, not the AI tool.

Last week I wrote about the internal conditions organisations need to execute AI innovation: SPOC - the systems, people, structure and culture that either enable or block it.

But before any of that work can land, there is another question, one that most organisations skip entirely: How well do you understand the ecosystem your AI strategy is sitting inside?

Kodak invented the digital camera in 1975. They held the patents, they had the engineers, they had the brand. But digital photography didn't displace film because the camera got better. It displaced film when the internet arrived and made sharing photos instant, free and social.

Blockbuster was offered Netflix for $50 million, dismissed them as a niche business and laughed them out of the room. Blockbuster had the distribution network, the brand, the customer relationships, and the capital to own streaming before streaming existed. What they couldn't see was the ecosystem shifting around them including broadband infrastructure, flat-rate billing, recommendation algorithms, the death of late fees as a cultural expectation.

BlackBerry held 50% of the US smartphone market in 2009 and lost it, and not to a better device, but to an ecosystem. A touchscreen interface people actually wanted to use. An App Store that gave developers a commercial reason to build for iOS. A media library users were already locked into. A consumer identity that made the iPhone something you wanted to own rather than something your IT department issued. 

These are all real examples of where the tool is rarely the disruption but the ecosystem assembled around the tool is.

Even further back in time, Henry Ford's customers wanted a ‘faster horse’. The problem was never the horse but it was the need to move faster than any horse could. The people who solved it didn't ask what customers wanted. They mapped what was becoming possible and built toward a destination their customers couldn't yet imagine.

That is exactly where we are with AI. We cannot yet fully define the problem it will solve, because we do not yet know what it will be capable of. Which means don't spend all your energy solving for what you know today.

Instead, stay close enough to the real human problem that when the capability arrives, you recognise the match. Map what exists. Watch what is being assembled. And never lose sight of what you are actually trying to solve.

What complementary assets does your AI strategy depend on that you do not yet control?