LLM’s are great until the entire internet has them in six months #productmanager #productmanagement #techtok #ai #strategy #product #engineering #mentor #mentorforu #mentorship #mentortiktok #career #techtoktips #careeradvice #explained #Tech #tipsandtricks #mentoring #chatgpt #startup

When everyone runs after something, it becomes commoditized really fast. That is happening right now for large language models. That means if you are planning in your road maps or if you're planning to build something with a large language model, the notional value that I assigned to the cost of that intelligence is zero. That doesn't mean it's actually zero, like it's still going to cost something. But the cost is decreasing because competition is flooding the space. If you think about it from a differentiation point of view, if everyone is building these chatbots, if everyone is building something with a large language model, pretty soon that's just going to become like the wallpaper. Either it works or it doesn't in your workflow, but either way, it doesn't differentiate. What does? I'm here to suggest two things we'll differentiate, and yes, it does include AI. It's relevant to AI, but it's not the LLM chatbot experience that everyone seems to be building. Number one, having a unique data source, having something that you can train your artificial intelligence app on that is not in the public domain. There's no way to replicate that. That's data that nobody else can get at. Number two, being really clear about the value of saving your user's time with better workflows. So improving discovery, improving actionability, whatever it is that is relevant to your users in their workflows, that is going to stand out not because it's impossible to replicate, but because when you combine it with the data and a unique piece of intelligence, you can get to workflow savings that you couldn't get to any other way. So that's a long way of saying. The LLM itself is going to feel commoditized and valueless in the next couple of years, because everyone will have actually launched all the things they're building, and it will be all over the web, and it will be done badly with far more than it's done well. But there will be cases where it's really transformative. In the cases where it's transformative, we're going to be associated with unique data sets. They're going to be associated with really thoughtful workflow placements that truly save a ton of time. And when you do that, you get breakthroughs that are hard to replicate, because people start to build habits around what's easy for them, and then you get that stickiness. So if you're thinking about how to put AI into place, think about the data, and think about the ease of use in the workflow, where can you put this in the workflow where it makes a 10X difference. That's my tip for you.