Based on Helena Liu interview with Thales Teixeira.
https://www.decoupling.co/about
Source:
https://youtu.be/wNUIhCI_jsw?si=ODLX9eHFd2xBpsSl
The hardest part of building something with AI isn’t the technology. The technology has never been easier to use. What’s hard is figuring out what to build.
This seems counterintuitive. We’re bombarded daily with news about AI capabilities – chatbots that can write essays, models that can generate images, agents that can navigate computers. The possibilities seem endless. But that’s exactly the problem. When everything seems possible, how do you decide what’s worth doing?
The answer is to focus on problems, not technology. And not just any problems – you need to find what business professors call “weak links” in the customer value chain. These are the points where customers are consistently unhappy with current solutions.
Take McDonald’s recent attempt at AI drive-throughs. They started with the technology – “Let’s use AI in our drive-throughs!” – rather than with a problem. Drive-through ordering wasn’t actually a major pain point for customers. The result? The AI misunderstood orders, created longer lines, and frustrated customers. They built a solution in search of a problem.
Contrast this with PayPal. They started with a clear customer problem: sending money online was slow, expensive, and complicated. Banks were terrible at it. PayPal built a solution that was faster, cheaper, and simpler. They didn’t try to replace banks entirely – you could keep your checking account and mortgage. They just took over the one thing banks did poorly.
This pattern shows up again and again in successful startups. Twitch didn’t try to become a game company. They noticed people wanted to watch others play games, just like kids visiting friends’ houses to watch them play. They made this one activity dramatically better – available 24/7, with unlimited choices of games and players to watch.
The best opportunities often come from activities that seem minor or obvious in retrospect. Before Twitch, who would have thought watching other people play video games could be a billion-dollar business? But that’s exactly why it worked. The established players ignored it because it seemed too small or silly.
So how do you find these opportunities? Start by mapping out everything customers do to get value from existing products or services. Which steps create value? Which ones just waste time? Which ones do customers consistently complain about?
You don’t need fancy market research. Start with problems you’ve experienced yourself. Talk to friends and family. Read online reviews. The goal is to find patterns – activities where lots of people are unhappy with current solutions.
Only after you’ve identified a clear problem should you think about how AI could help solve it. Maybe it’s using generative AI to create content faster. Maybe it’s using predictive AI to make better recommendations. Maybe it’s using automation AI to eliminate tedious manual steps.
The technology itself doesn’t matter as much as you might think. Netflix didn’t start with amazing recommendation algorithms. They started with a clear problem – video rental stores were inconvenient – and gradually improved their technology over time.
This is good news for anyone starting out. You don’t need cutting-edge AI capabilities or massive resources. You just need to find a real problem and solve it better than existing solutions. The technology is just a tool to help you do that.
The hard part isn’t building the solution. It’s finding the right problem to solve. Focus there first, and the rest will follow.
Helena Liu is one of my favourite youtubers, passionate about AI and automation.
Thales Teixeira is a prominent academic and consultant specializing in digital disruption and marketing strategies. As of now, he holds the position of co-founder at Decoupling.co, a consultancy that focuses on helping established firms navigate the challenges of digital transformation.
