Technology comes in waves. Our industry is always awash with excitement about something. I’ve watched these waves come and go over the last thirty years. At one point, and for younger developers who weren’t around at the time this might seem hard to believe, there were conferences around XML. About how XML, and schemas, and ontologies, would change everything. We had conferences and excitement — as the iPhone really did change everything — around location, discussions about how the world would change with the availability of location all the time. There was Big Data, and the arrival of NoSQL databases. There was the Internet of Things. Machine Learning. Deep Learning. Edge computing, because sometimes things come back, and we just change the name. Voice. Then there was crypto, and NFTs, because sometimes the ideas aren’t so good and add nothing, but we get excited albeit somewhat briefly about the technology anyway. Right now there is AI.
As engineers, developers, technologists, what we are excited about is the technology itself. We are tool users who have gained access to to a new tool, a new lever on the world. We frantically run around and try and use the lever to pry things apart, things that sometimes we weren’t able to open before. But sometimes we just use the new lever in the same way we did the old, or worse, misuse it as a hammer, and beat a square-shaped technology into a round-shaped hole.
But the people we build our tools, services, and products for don’t care about the technology we’re excited about. They care about their problems; they care about solving those problems. One of the best ways to figure out — as a potential user and customer — if a product will solve your problem, is to consider whether the company talks about your experience first, and about how the tool you’re thinking about using solves the problem. If they lead with what technology they use, not with the problem that they’re solving, you should walk away.
You shouldn’t have to market your startup as “doing X with Y”. Just “doing X” should be enough, and to smart investors it’s a huge red flag that you’re just searching around for something to solve with that new lever.
This tendency, our tendency as an industry, to focus on the technology rather than the problem we’re trying to solve, leads to some overly complicated solutions. Putting aside the absolute failure of crypto to find any product-market-fit, and any good example of where “with blockchain” added value, one of the really good examples of this is the Internet of Things.
The idea of smart objects in the home was greeted with a lot of enthusiasm inside our industry, but out in world people with homes, and children, and pets, didn’t see the benefits. Adding smart lightbulbs to a room added friction to day-to-day interactions. Held up as an success story they were, and still are, anything but.
Light bulbs are things that you can turn on and off with a switch on the wall. You can make a smart light bulb unresponsive by using the light switch — or by anyone else in the house using the same switch. A smart light bulb replaces the thing we use every day, the light switch, but it does it poorly. Really, we need to replace the switch, not the bulb.
The problem there is of course that the light switch needs an electrician to install it, and a light bulb doesn’t. With smart light bulbs, you’re trading one-time friction at installation for friction every time you turn the bulb on or off. That’s not a good trade, and in general it’s one that users weren’t willing to make.
As an industry, we shipped the thing that was easy to adopt rather than a thing that actually works. We shipped the thing that we believed was the fastest way to those exponential user growth curves that every startup needs, rather than the solution to the users’ problem. It’s a mistake we make a lot.
Of course, sometimes the technology matters a lot. New technology, new levers, are fundamental to solving problems. The iPhone really did require large capacitance touch screens, and a sufficiently fast data connection, to change everything. It couldn’t do it without them. So the questions is, what distinguishes genuine enabling technology from cargo-cult invocations of the latest trend?
What made the iPhone different from smart bulbs? The touchscreen wasn’t bolted onto the existing phone experience: it dissolved the old interaction model entirely. You weren’t learning a new way to do the old thing, you were doing something that felt like nothing you’d done before. It wasn’t a phone, it was a computer that looked like one. The technology became invisible precisely because it was so fundamental. Does the new technology disappear into the experience or does it sit on top of it, demanding attention?
We have to ask that question of our latest industry obsession. Where does AI disappear into genuinely solving problems that weren’t solvable before: protein folding, real-time translation at scale, making programming accessible to people who think in natural language? And where is AI the equivalent of the smart lightbulb: chatbots that add steps to support flows, generated content that requires more editing than writing? Where is it used as a badge rather than a capability?
As an investor I have to then ask, when a founder says “we’re doing X with AI,” which kind of AI is it? Is it an iPhone, or a light bulb?