In Search of the AI Cannonball

By Daniel Horn (Chief Technology Officer)

Almost every day, I get a question something like this:

“Hey Daniel, have you heard of this new AI product called XYZ? It has this crazy new feature I’ve never seen before! How do you think it’s going to change the insurance industry?”

I hear this so often that I’m running out of ways to say, “I don’t know, we’ll have to wait and see.” But the question is valid, and even among the big brands in the AI space, there’s a lot to sift through. We all know the names by now. ChatGPT. Perplexity. Anthropic. Gemini. Who’s going to win? Which one is going to offer the best returns for consumers and businesses over the long term? Which one do I pick for the problem I’m trying to solve right now? 

And more new names pop up every day. Crunchbase reports that AI startups received $100 billion in venture capital in 2024, or about one third of all venture dollars. You read that right. One out of every three venture capital dollars in 2024 went to an AI startup, an astounding number. Clearly, most of these companies are going to fail. So what’s going on? How do we make sense of all this?

Here’s a framework I’ve found useful as I watch the development of the AI space. In their 2011 book Great by Choice, Jim Collins and Morten Hansen wrote, “fire bullets, then cannonballs.” They meant that companies should run a lot of low risk, low cost tests (bullets) to see what works before jumping all the way into a new product or venture (cannonballs). And that’s what we’re seeing with AI. The free market and venture capital are financing crazy amounts of bullets, because no one knows which tool or which breakthrough is going to make the biggest difference, whether in the insurance industry or elsewhere.

There are still a lot of outstanding questions. How will insurance agents continue to protect client privacy while improving client experience? How can we ensure that relationships keep their rightful place as the most important part of any business, while still innovating? 

But at least we have a framework to understand what’s going on. All the venture money, all the startups, all the seemingly parallel language models – they’re just bullets in search of a cannonball. Like BlackBerry before iPhone, we’re probably going to have to watch a lot of AI startups peak and fall before we see the big hits. But we will see the big hits. Will we have to sit through a full AI boom/bust cycle that mirrors the dotcom cycle of the 90s? No one knows, but I also don’t think it would surprise anyone. AI will change the world. But for now, the free market is just firing bullets.

Until someone finds the AI cannonball.