From Horizons to Frontiers: Rethinking the Possible

The Three Horizons model has served innovation teams well for decades. (I'm a big fan). But when it comes to AI, it feels like there's an opportunity to evolve it even further.

AI isn’t just another technology that trickles through the usual phases of efficiency, expansion, and transformation. It collapses those boundaries entirely. Things that once lived in the “impossible” box can be viable overnight by a single new model release. Why does this matter? Because it's never been faster, easier or cheaper for a new entrant to undermine your market share.

So rather than talk about horizons, should we think about frontiers - the edges of possibility that shift every time a new model drops.

What’s exciting - and destabilising - is that the boundary between these frontiers keeps moving. As new models land, fresh use cases suddenly become possible. When Google’s “nano-banana” was released, people chatted about an 'unlock' score: how many genuinely new use cases were enabled, not just improved.

1. The Efficiency Frontier – doing what we do, faster

This is where most organisation are right now. They're automating the manual, speeding up analysis, tightening workflows. The use cases don't change. It’s where AI helps you reduce resource and/or speed up....and, ideally, frees up some headspace for bigger thinking.

2. The Expansion Frontier – doing what we do, differently

Here’s where I think things start to get interesting. AI doesn’t just optimise existing use cases, it allows you to explore new ones - whether it's a stretch into adjacent markets, and experiment with new forms of customer experience. The paradigms don't change, but your organisation can lean into areas that were previously out of reach.

3. The Impossibility Frontier – doing what we couldn’t do before

When we were working with customer service teams at a large European-wide organisation, they were too busy to follow the ins and outs of latest tech developments. So when we needed them to imagine the future, we did a simple exercise: we handed them a magic wand.

The Impossibility Frontier is that magic-wand space. It's the stuff that used to be too hard, too expensive or slow to be viable, or simply beyond the realms of possibility. But with each model drop, new possibilities emerge - for everyone. Think hyper-personalisation at scale, real-time co-creation, or completely new interaction models.

We're working on a simple toolkit inspired by frameworks like Where to Play, to help organisations explore that second and third frontier - the “previously impossible” space. It combines expansive thinking exercises with AI exploration prompts to hunt for those magic-wand opportunities.

We’ll be testing it out soon, so if you fancy stretching your imagination (and your use cases), keep an eye out or message me.