Since the launch of GPT-4 a year ago, I’ve seen developers and AI engineers talk about building agents that autonomously do tasks for you all over YouTube. As is the way these things go, they were super clunky and hardly worked at first. All the installs were super manual, and you had to sift through a sea of documentation to get them to work.

See, the thinking here is that agents mean autonomy, and autonomy leads to AGI. So, for a long time now, there’s been a focus on building agents that actually work.

Slowly but surely, though, these agents were able to do more things accurately and act as developers, program managers, or even executives to build entire applications from scratch. Following Microsoft Build this week, they are now being directly integrated into laptops like the new Copilot+ PC Surface Laptop and the new Copilot Studio.

It’s a fun time to be a dev, actually. I don’t think this has to be the end for developers. I think these are only baby steps, and there’s still much to be done.

Personally, I hate overhyping AI and its capabilities, but I paid close attention this week during Microsoft Build (aka, I stayed home and watched all the demos on YouTube). Let me tell you, this Copilot and Azure AI stuff changes a ton (for devs).

But there’s definitely something more brewing at OpenAI besides a ScarJo scandal and NDA drama. Especially with the latest model, GPT-4o, now being offered for free to all users. I’m just glad that, for a while, everyone now has free access to the best model available. Personally, I’m still mostly using Claude 3 Opus as my daily driver.

Why Does Any of This Matter?

When the developers get new toys, it usually means more new toys for the public, and while devs have been using Copilot Studio to make work faster and easier for teams for months, now these tools are generally available. This means you could start seeing more businesses tinkering with agents that finally actually work.

More devs playing means more data gathered from users, which means more powerful ways to siphon and extract from users. Yes, that’s real. It’s all just a game of who can collect the most data on you. But you already knew that.

It’s also already made life easier for help desk workers (my first “tech” job was tech support), and it seems with each passing day, another role gets upgraded.

Along with a slew of AI products shipped this week was a feature called Copilot Recall. No, not Total Recall… not yet, anyway. But we’ll probably get there someday.

While Recall looks like magic in the demo you’ve likely seen by now, as a former engineer, it’s not exactly magic at all. Here’s how you can build with Recall today. It’s essentially an interactive, searchable Internet Archive but for your laptop. It takes a snapshot of your screen every 3-5 seconds and holds it for a period of time. You can then search or scroll through your history to find and interact with whatever it is. The thing is, applications must enable it first. That’s why, in the demo, you’ll notice Discord set up in the browser rather than in the standalone app.

Not only does it take up storage space, and though Microsoft does have a hardcoded privacy and security setup, there are reasons to be concerned. Sure, maybe Microsoft isn’t collecting the data since Recall is stored locally, but what happens if you use a PC that has it turned on and doesn’t alert you?

It’s giving trust issues.

Also, it’s only available on these very specific new Windows Copilot+ PCs because they have a new NPU chip and on-device SLM (small language model) called Phi3. So, if you’re on a Macbook, like I imagine most of you reading this are, I don’t think you have anything to worry about just yet.

But that’s just one of the things shipped this week at MS Build. Things like upgrades to Power Platform for building low-code apps and updates to Azure AI. One of the things I continue to be amazed by is the advancements in accessibility tech. Whether it’s making general UI improvements or completely new products to make using tech more accessible for all, that’s the hope here.

The Internet is Evolving, and You Must Stay Sane

For the past couple of weeks in the Boys Club discord, I’ve been tossing around this idea of how the New Internet truly is evolving in front of us. It’s not a joke anymore or some great big wish. Even if we, the users, are often making a fool of ourselves (see: “scenecoins,” memecoins, AI girlfriends).

I thought I’d let you in on my half-baked ideas and see what you think too.

I think it’s time to acknowledge that the internet we know and haplessly love to despise today will look nothing like itself in five years’ time. I, for one, am relieved to know and observe this shift. We deserve a New Internet, one which could look like a total rebirth or a reimagining of how we interact with data, which is essentially what the Internet is to all the people funding it. For the sake of those people, I don’t think they’ll get their way and continue to extract while not providing outsized value.

In the meantime, though, the internet we have now is experiencing some growing pains. All as expected with new disruptions.

Consider this: In this week’s presentation, before a talk on stage with Sam Altman, Microsoft CTO Kevin Scott said that their team has previously deployed 2 huge supercomputing systems for OpenAI with a recent third the scale of a whale when compared to the orca and shark size of GPT4 and GPT3 respectively. Here’s the talk so you can listen for yourself. He noted that “the platform is becoming so much more capable and so much cheaper over time that everything you have in your imagination that’s too expensive to do right now is going to be cheap and robust before you can even blink…” So clearly, we have some massive exponential growth going on.

This all amounts to incremental nudges of an entire platform shift—as in rethinking what it means to be online at all. Of course, it’s a whole lot more complicated than that. But this New Internet is in the air, and as this Overton window shifts, we, the users, have the opening we need to build meaningful change.

And I don’t mean we’ll all switch to being exclusively onchain in 5 years. Maybe, but at this rate, I’m still a bit skeptical. At least not until the blockchain space can actually get its act together, which doesn’t seem likely anytime soon.

I believe that as these radical shifts in interfaces and information progress, one technology must die before another slowly takes its place. Just as AOL, Netscape, Windows Phones, etc. slowly died off, so too will various bits of the Internet fall away to create new, more dynamic, interactive interfaces for wisdom gathering and social connection in their place.

This is the New Internet moment. I don’t agree with everything Sam says, but when he compares this time period to the mobile era of 2008/2009, he’s not wrong.

One thing I hope to see as these shifts happen is new social contracts, new social networks, and even more personalized ways of connection that allow more creative freedom. As our social lives blur the physical and digital, we’ll be afforded new opportunities for discovering new communities and, most importantly, new people along with their creative minds.

Will Your AI Agent Match My Freak

Right now, all of our circles are so tiny. Even if we feel like our spaces make an impact on culture, it’s actually quite small compared to what we can do within communities of tomorrow. For example, here’s Whitney Wolfe Herd talking about recent nudges in the dating world toward personal AI dating agents on Bumble that might go on dates for you in advance to measure chemistry between potential partners.

The first dating app I ever used was OkCupid, circa 2008. If we're not careful, the dating apps of tomorrow could have even more features that keep us apart or even venture into Black Mirror’s “Hang the DJ” territory.

However, I have some hope that the discovery of new people and new communities and the proliferation of AI-everywhere devices will actually help us connect much more deeply than we could imagine today. I believe we’re waking up to that very dire need.

You’ll see more of this Internet Abstraction over time that pulls you away from the time-consuming frustrations of finding a match, like Google working on doing your shopping for you rather than the friction of returning or choosing the right item.

Soon, you’ll tell a device what you need, and it’ll be so personalized that it delivers exactly what you want. Why bother searching when the systems you use already know what you need?

At this point, I don't trust Google to deliver what I asked for, seeing as how they recently had to remove the experimental AI Overview feature from searches, just as countless news outlets like The Verge tell users how to turn the feature off due to inaccuracies.

Because of this platform shift, it's likely that more users will shift their social media usage as apps change, die off, or we simply get collectively fed up with how we’ve been slowly lobotomized over time.

The Next Battleground is the Mind

As the New Internet is born, the next space where humans will be challenged is the mind and the impending identity crisis.

We’re already being tested by a Metacrisis: a crisis of meaning, loneliness, and purposelessness. A few months ago, I watched a talk on this topic with Ian McGilcrist and John Vervaeke, moderated by Daniel Schmachtenberger. This is one side of the mind battle, but it’s far more multifaceted than this. Going beyond AI-generated misinformation and deep fakes, we’ll all be faced with challenges of the mind and purpose.

As people around you continue to lose their jobs and fail to get new work or take up the tools to do better work, they’ll be faced with a new set of challenges similar to the ones we see in any of the other Industrial Revolutions.

Every day, philosophers, spiritualists, and scientists unravel new information about consciousness and the mind that pushes the idea of human experience further to the edge. I’m sure, at some point, this will lead to concurrent realizations in the computing and robotics fields as they converge.

No matter how cheesy it sounds, it might be time to ask yourself, the passionate generalist, a few questions. I’ve been doing this over the past few weeks while reflecting on this writing project.

What do I actually want to be experiencing?

What career do I want?

What is my ideal workday?

Can I actually let go of the fear?

I’ve said this many times before, as the world around us expands, the place to look is inward. Personally, I’ve been frustrated lately by looking for the looker in my practice. It’s for this reason that I believe discovery, awareness, and their interaction with embedded AI are the next frontiers to explore.