Yesterday, I got a message from a friend in one of our programming group chats.
“Yo! I just finished watching the Microsoft Build event. They announced some really cool stuff. They’re going to integrate OpenClaw with Windows and allow you to run AI locally. You guys should check it out.”
And then he sent a link to the keynote.
Now, while normally I just wait for Fireship to make a video about it, like most sane programmers, I thought... huh, you know what this would make for a decent email. Maybe I can report a little bit about it and tell all of you lovely people about it!
So I actually skimmed the 3-hour presentation, read a few summaries, and saw what people were more or less excited about the announcements. All to give you THE SCOOP of what’s happening on the Microsoft side of tech and what to watch for.
And in that process, I’ve come to realize something.
I’m sorry, but they are just SO BORING 😭
I find that unless I pay very close attention to every word said, it’s easy for me to lose track of the meaning behind the words. I mean, look at this sentence, which was said in the conference:
“The second salient point is that there is a new operating point at the frontier where you can use a very efficient reasoning model and a coding model and achieve frontier-level performance because you’ve done the hard work of creating that environment, that hill climbing machine in which these models with your traces can hill climb to the frontier.”
Like bro, what does that MEAN?! What does frontier mean anymore?
But I digress. Whatever my opinions are, I still gotta do my self-appointed job. So, I’m going to make this as easy to understand as I can for you guys.
Microsoft is pushing to allow agents to be used everywhere. Locally on your own computer, on the cloud, on the edge, uhh... off the edge?
They are trying to create what they call a “Frontier intelligence system”. That means go agent-first, everywhere. Imagine having access to artificial intelligence and agents on every device, on the go, working for you.
So with that goal in mind, to do that, they have announced:
A new, high-performance developer device called the Surface RTX Spark, designed to run local models.
Project Solara, a new agent platform for “agent-first” devices. Allow agents to work across your devices.
Windows 365 for Agents, which allows agents access to cloud PCs for use inside any business environment.
A bunch of new AI models called the “MAI model family”. Better, faster, stronger… all the normal stuff you hear when a new model comes out.
That’s most of the interesting news around everything agent-focused. If you didn’t find that interesting, there is some more dev stuff that is with a doubt positive, like how...
Y’all, you’ll be able to use Grep on Windows soon. You hear that?!? They’re bringing in over 70+ Linux utils like curl, sudo, touch, head & tail. As well as support for WSL containers and even native support for tools like Starship, Zsh (Zshell), and Homebrew. Hot diggity DAWG! I’m sure that’s a relief to a lot of devs. A welcome change.
And finally, probably the news that got ME the most hyped, which they ended the day on...
Microsoft announced the advancement of its Myana 2 chip, a new quantum computing chip that is 1000x more reliable than its predecessor. (It also now contains LEAD! Deadly, but effective.)
It’s a huge leap compared to what they had before. A lot of this progress is actually thanks to AI, which has allowed them to be able to cut their projected roadmap to building a quantum computer in half. They are now aiming to reach their goal by 2029. Which is personally crazy to me.
I’m not going to pretend I understand these advances on a deep level, and I question if they can actually do something like that in such a short time span. But at the very least... it does seem to suggest we’re not too far off from that possibility.
I’m not exactly sure if my body could take on another “AI-like bubble”, but instead with quotum computers 😓
Anyway, that’s it!
As for what I thought about it in general?
To be honest... having such a loud, constant focus in our industry around agents (in the same way we had around AI just a few years ago) reminds me of an old saying.
“If your only tool is a hammer, every problem looks like a nail.”
Even though we’re making access to this tool more open and easier to use, I do fear it tends to lead to an over-reliance on it for tasks we shouldn’t allow them to take. Like thinking, writing, understanding tradeoffs, or plain decision-making.
It’s an easy trap to fall into. It’s just far easier to write a prompt and get a response rather than do the hard, slower thing, which is to think and plan for yourself.
For example, when I wrote my “How to become an irreplaceable frontend dev” book, I purposely did not allow AI to write a single word of it. I spend weeks thinking, drafting, writing, and editing every part of it. 100% just me. Why?
Because I‘m a masochist I know it’s easy to “smell” when something was written with AI. When you sense that something is written with AI, you always have that nagging feeling… “Is this just the AI talking? Or is this the actual author’s thoughts?” And so you value it less.
Which I did NOT want for a book I wrote on teaching you to become obsessed with value. So funny enough, writing the book sort of became a meta-lesson on what the book taught. Which is cool!
If you’d love to get a free copy, you get one whenever you sign up to Exceptional Frontend.
- Mauro Accorinti
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.