Jay Cuthrell surveys AI trends including agents, edge computing, and on-device models replacing cloud-based systems. He notes remote sensing and robotics are growing, and predicts a breakout personal AI product could shift the market like email once did.
Chapters
Alright, it is time for the What the Fudge for March 22nd, and I've decided that I'm going to probably build out a little bit more of a mobile solution. I will probably get a dedicated microphone or something to that effect, something that I could use with the iPhone. I don't know that I'm gonna have a laptop with me all the time while I'm traveling, but the whole idea is to get this in and and get it done with the minimal of equipment required, something that would fit into a backpack. So a couple of things that I'll cover for this update, and it's getting late because I'm going to be attending the All Things AI event here in Raleigh-Durham, North Carolina. Really excited to see folks. I remember fondly last year, it was a really great event, and so I'm fortunate to join all of my peers from last year as well as some some folks from the NC Tech organization that should be in attendance. So I'm looking forward to that. And, you know, on that topic, um, AI, the notion of agents and coding has really been a large part of the news. As I segue into the actual podcast, I think of it as one of the elements around engineering an outcome is the focus on the harness. So that could be whether you're dealing with the model, the feedback, the team constraints, constraints. It's not, it's not like everyone is talking about new models all the time. So if you remember about a year ago, everything's going to be model, model, model. And now what you're looking for is people are doing things with said models. And some of that comes from where maybe a year ago, a year and a half ago, you probably would have heard prompt engineering. Now you're hearing more about context engineering, which is really trying to get folks to pull together Who is the person that's asking for what they want? What's their role? What's the, what's the goal? And then it's the meta or, or, or higher level abstraction of getting away from just pure prompt engineering. Then there's going to be this rapid progression of models. There's going to be probably an increase in the interest of putting models closer towards devices, closest towards the edges of the network. And that ties into what I believe will be this growth in remote sensing, telemetry, robotics, far more autonomous systems that are taking inputs in from other places and putting that together and synthesizing that information for the benefit of a robotic or autonomous outcome. Now, some of this is most likely going to be very, very large companies that have capitalized not only on the investment of AI infrastructure, but probably also have some, some amount of investments in robotics, etc. So probably won't be shocking to see companies like, for example, Amazon showing up in those places. But I think what might be interesting is smaller players doing things in very niche areas or very specialized areas. And so I would, I would keep an eye on that. And the only other thing I was going to mention was the notion of where all of these investments are beginning to play out and where they might show up next. And if you play this out a little bit further, what might happen— markets are a little interesting right now in the sense that we live in interesting times, which I think was— someone said that's, that's a type of a curse, may you live in interesting times. But I think of the investments right now as a combination of AI, automation, roboticization, and the question of the dystopian fork in the road? Does, does, does the cliché of the Skynet moment get ever closer because of the things we're doing? Or is it that it's only in the news or in the coverage when it is dystopian? And so I think an important element here is to also look at the amazing progressions that are going to be taking in, taking in these, these advances and applying to things like robotic surgery or assisted robotic surgery, or potentially even, um, what, what, what would be the outcome of bringing all these disciplines together and solving some of the most challenging problems? Because, uh, robots can continue and run and, uh, don't take days off. They could certainly, uh, need repairs, but, um, I think you, you'll even see a, uh, situation where robotics will come into more places in smaller settings. And I think of this somewhat in the same way that I thought of the early introduction of 3D printing, or so-called additive manufacturing. Now, 3D printing was a hobbyist type of approach, but when you look at what's happening in that industry, there is this race to lower, lower costs. And if you follow the progression of how inexpensive, relatively speaking, flat panel displays are for home TV purposes. The, the, uh, the, the, the price just seems to get lower and lower while the size of the panels get larger and larger and larger. And so one would expect that some of that would also probably take place in robotization, combinations of robotization with smaller language models placed into the realm of the near phenomenon of where that robotic system is taking its cues, again, with a goal or an outcome. And so these don't necessarily they don't have to be very large robots. They could be much smaller robots, possibly even tabletop. And so it'll be interesting to see how the combination or confluence of all of these things and the convergence of these things come together over the next 2 to 3 years. But if you're still looking for where things might go next, I would say seeing other participants beyond, say, the coworker Claude Anthropic stories you've heard, now expanding to what that means for Google, with its personal AI assistants expanding on its Gemini. It's sort of, if you remember how important Gmail was as compared to some of the other email providers at the time, if you can think that far back. Now imagine what might happen if the same way we think of email and what Gmail did for it. I'm not saying Google will do this, but what if, what if Google's personal AI assistant in Gemini was just good enough to really catch hold and then everyone starts using it in the same way that OpenAI's ChatGPT was? Or the way that Anthropic Claude is today, or even Perplexity or some of the other earlier players. What happens when there is one breakout? Someone in the product engineering side really figures this stuff out, and now you have an entire generation of new net users that are pulling all these stories, all these concepts, all of these things that they would use to get their jobs done together. Now imagine that not necessarily having to go all the way over the internet to another data center filled with AI infrastructure, Imagine a lot of that running locally on the device itself. Locally on the device. Truly private compute. Imagine that secure private compute concept on a device. And so this is where, if you've been following some of the Apple coverage of Apple Intelligence, the, the Apple definition of AI, it'll be fascinating to see if that happens. Of course, obviously Google has its abilities with its own Pixel line of phones, and certainly There's been other phone companies that have been shipping hardware capable of running a very small language, um, uh, large language model or a small language model, I should say, on those devices. So it'll be interesting to see how the edge of the network becomes increasingly AI without having to necessarily rely on a network and, uh, a larger network of more generalized AI infrastructure. What if the, what if the AI is literally baked into the phone, into the device, into the robot? And with that, I'm going to call it a wrap and plan to get a more mobile version of this podcast recording setup. Maybe, maybe in a few weeks, we'll see. But I'm going to try to keep the podcast Sundays. Okay, take care. Again, this was What the Fudge for Sunday, March 22nd, 2026.
