You Haven't Had Your AI Moment Yet
The future is agentic, whether you like it or not.
Months ago, I got a taste of what the agentic future and freedom means when I was able to automate a few tasks with OpenClaw. But I have to admit that it took a bunch of time on maintaining and kept breaking or running into permission issues. Then on September 3, my world changed when Instinct launched and I got early access. I had my transformational experience. By that I mean the experience that shows you the potential of the agentic AI future, where an agent finishes a task for you that you just did not believe a computer could do on its own.
Increasingly, I’m seeing the world divided into two groups: The first group has not experienced personal agents doing some fantastic, magical thing for them. When you ask them about AI, their responses range from “Oh, it’s great, it’s helping me write emails” to “yes, I definitely think it’s the future.” But you can see they’re mostly repeating the words they think they should be saying vs speaking from first principles, or from their personal experience.
Then there’s the second group, who have gone through a transformational experience where personal agents have completed tasks for them that wowed them, that showed them the potential of agentic AI. These people are not same since that day, their perception of this technology has changed, along with their belief in what it can do.
For me, that happened when I first got early access to Instinct. Even though I had fairly stable automation already set up with ChatGPT, Instinct showed what it actually means to delegate. I sent it tasks as easy as ordering me a burrito to as complex as figuring out why the auto pay on my water bill is not working & getting refunds for wrong charges on my account. In each instance, Instinct performed amazing, sometimes needing guidance and sometimes making minor mistakes but it was site to behold that I told it what i needed and it got on it relentlessly, like a really motivated type-A employee. Whatever it lacked in skill if at all, it made up in persistence. Instinct since then runs my chief of staff email inbox and help manage scheduling for everything. His name is “Mr. George” and I asked it say hi to you guys so here is Mr. George’s message for you all.
Then came Muse. Muse took the world by storm. I realize that I am in a tech bubble, but at least my tech bubble was taken by storm. Muse took what Instinct started to another level. People whose imagination had died down because Instinct had no interface and just lived in your iMessage/WhatsApp were finally able to see an app, think a little differently, try new tasks. Meta also nailed the app execution so well that it became clear to people that they can trust this app (vs plain old Meta reputation), they can store their credentials, and get things done. You could watch the browser working as Muse went in to do stuff for you and i believe, a lot of trust got built because of what people could see.
Muse completely transformed many folks. It even transformed me. I remember a particular day when Instinct was down for a few hours and I switched to Muse so seamlessly, even gave it my chief of staff email account to run, that I questioned: do agents have a moat at all? Or do we just see them as a means to an end? For me, it didn’t matter what agent it was, as long as it was finishing the task. Place the pickup order for food after finding which Chipotle is on my way home, out of diapers - go figure out my Coterie subscription, unknown charge on my credit card - pls go get to the bottom of it. I was a now a junkie who had tasted freedom from the mundane tasks. After these, I tried more personal agents, namely Fo, Grok Bot, Dots, Claude cowork, Wabi, and Hark. It got so amazing that i have had more dr appointments in the last 1.5 month than in my whole life, because calling/scheduling was my kryptonite and agents just went ahead and did it. Shout out for Fo, who even makes outbound calls for you.
got so bad that wife made a meme of me in family chat. i had to start watching my screentime with agents now.
In this whole thing, one thing became clear: the moat was no longer the interface. The moat was the quality of the responses. I started noticing the agents that tried to be too clever or did not give thorough answers or the things they were good/bad at. For example: sometimes Instinct told me, “Yes, I’ve added this to your calendar, but I cannot add the drive blocks because I don’t know the address.” And I told it to figure it out, and then it figured it out. With Fo, I struggled as it started signing each email with Fo, I liked the team’s growth strategy but I think in the personal agentic world this kind of thing counter signals, and I stopped using it for any email delegation. Wabi’s categorization of life’s topics was very good but for me, iMessage interface has always been the vision. Even before instinct, that’s what my moonlight project was as I and my friend were making a general purpose AI chatbot on text/iMessage, but then GPT launched and we lost steam. I know I am kicking myself too, should’ve stuck to it.
Let me stop with the reviews, I’m not a product reviewer. That’s not what I am good at. I’m good at my own commentary and perspective on things. So here it goes.
The future is agentic, whether you like it or not. Similar to how we have a lot of apps today, we will have a bunch of agents. Maybe you will have a work agent, a home agent, a car agent, a food agent. But we will have agents to finish a lot of tasks, and maybe one general-purpose agent that wins and takes over, for all your life tasks.
Now what do I mean by an agent? Think about how an app works today. You open it, you search, you tap around, you check out. The app is just a tool and you are the one doing all the work. An agent is different, you tell it what you want and it goes and does the work for you. When I say order me a burrito, I am not opening any app. The agent finds the Chipotle on my way home, places the order and tells me when it’s done. Same with the water bill, I did not log in anywhere, I just said figure out why the auto pay is not working. So you go from doing the task to giving the task. And once you have given a task away like that, it is very hard to go back to doing it yourself.
I think the future is agentic because humans always find the path of least resistance in doing things. So once these tools are available and people can offload tasks that they were anyway doing as a compensating behavior, they will. One example I use is the day we figure out how to keep our teeth clean and healthy without brushing, nobody will brush their teeth. Brushing is basically a compensating behavior towards an outcome that we like. Same with ordering the food, fixing the water bill, replenishing the diapers. Nobody wants the task, we want the outcome.
The future is agentic, whether you like it or not. Similar to how we have a lot of apps today, we will have a bunch of agents. Maybe you will have a security agent, a home agent, a car agent, a food agent. But we will have agents to finish a lot of tasks, and maybe one general-purpose agent that wins and takes over, kind of like your home automation and stuff.
Now what do I mean by an agent? Think about how an app works today. You open it, you search, you tap around, you check out. The app is just a tool and you are the one doing all the work. An agent is different, you tell it what you want and it goes and does the work for you. When I say order me a burrito, I am not opening any app. The agent finds the Chipotle on my way home, places the order and tells me when it’s done. Same with the water bill, I did not log in anywhere, I just said figure out why the auto pay is not working. So you go from doing the task to giving the task. And once you have given a task away like that, it is very hard to go back to doing it yourself.
I think the future is agentic because humans always find the path of least resistance in doing something. So once these tools are available and people can offload tasks that they were anyway doing as a compensating behavior, they will. One example I use is the day we figure out how to keep our teeth clean and healthy without brushing, nobody will brush their teeth. Brushing is basically a compensating behavior towards an outcome that we like. Same with ordering the food, fixing the water bill, replenishing the diapers. Nobody wants the task, we want the outcome.
My second prediction is that the interface is not going to be relevant and UIs will be dynamic. Let me tell you a story. When Steve Jobs showed the iPhone next to the Blackberry, he said, look at the Blackberry keyboard.
All the keys are there whether you use them or not, even when you’re just looking at the screen. And one of his promises was that with iPhone, you don’t have that problem. When you don’t need the keyboard, it’s just gone. Similarly, we are stuck in that Blackberry keyboard era where all the UI is always visible, and we are going to a point where only the UI that you need is generated on the fly by AI to accomplish your goal. It is going to be beautiful. It’s going to be the future.
Let me make this simple. Open any app on your phone right now, say your bank app. There are tabs, menus, settings, maybe 20 buttons, and you came there to do one thing, like checking if a payment went through.
All of that is sitting on the screen whether you need it or not, same as those Blackberry keys. Now imagine you just ask, did my water bill payment go through? And what comes back is only that answer, maybe with few buttons that fit the use-case, and nothing else on the screen. Tomorrow you ask a different thing and you get a different screen, made just for that question generated on the fly. That is what I mean when I say the interface is not going to be relevant as UI will not be fixed, it will change for each use-case. You will not have to learn an app anymore, it will show up in the shape of what you asked for.
So if you haven’t tried personal agents, ping me and i am happy to share some invites. Don’t sleep on this, you’re going to be left behind.
I love you all. Keep going.
-Aakash
https://x.com/aakashpathak_
Originally published on Substack.