Era of Agents: Your Device Is Your Personal Assistant

Hey guys. I’ve been talking for a while about the fact that we’re entering an era of agents, and that our phones and personal devices in general will become our personal assistants. And for that to happen, what we need is really one thing: to be able to delegate tasks to this personal assistant in a convenient way. So first of all, it should be convenient for me — it should be able to capture my intent, actually understand it well; and second, it should be able to complete it on my behalf.

And we’re kind of getting there. Someone can claim that we’re already there; others will claim that it’s not good enough. In my opinion, we’re kind of in between. What we have today is a bunch of hacks that make it good enough, but there’s still a lot of room for improvement. Recently there was news about big labs working on interactive models — models that actually react in real time to input, rather than being turn-based models. And Apple works on their Apple Intelligence to make it happen on the infrastructure level: to allow the LLM to complete tasks on your behalf on the phone, search the web, open apps, and do some stuff.

So what does it mean? What will change once we’re actually there? In my opinion, what will change is that as this assistant becomes more and more capable and powerful, we will open our phones and touch their screens less and less. And what that means is that this brick I’m holding right now in my palm is actually likely going to disappear — because I will not need the screen. I will not need to carry it all the time in my pocket, and remember not to leave it in my car or in my kitchen when I leave the house.

And the funniest part is that when Steve Jobs introduced the iPhone, he emphasized multiple times that the iPhone is actually a combination of a phone, a calculator, an internet browser, and whatever — consolidating all these functions into one device. And what is going to happen now is actually the opposite, because we don’t really need this single, centralized way to interact with technology. The phone will kind of fall apart. It will go into some kind of smart glasses. It will go into smart pods, maybe smart watches, some kind of smart wristbands — or other devices; maybe it will be integrated into clothes or shoes. Basically, we’re moving into a modular architecture of devices, where I can wear any combination of these devices. They will figure out how to interact between themselves, work together, and provide me a seamless experience — and be the interface for interacting with my personal assistant.

And what that means — and that’s a very nice, bright future I’m looking forward to — but what it means, and what’s important at least for me, is that the way we interact with the web, the way we share knowledge, the way we source knowledge will completely change. The web will become agent-first. We will build APIs, websites, and whatever, first of all for agents rather than for humans — because humans will not be interacting with this directly. And that’s actually the topic that fascinates me and doesn’t let me sleep at night. This is a topic we’ll try to explore in the next video. See you guys, and enjoy your weekend.