Loading…

Issue 338: An earth rattling change in software

The world of software development has always been about big booms and tectonic shifts in the industry. You know as well as we do that AI has been one of those cultural shakers. Even now, after the first wave of changes have rocked our technical worlds, we’re still feeling aftershocks. This week is all about those smaller shifts and what they mean for software. Live from Microsoft Build, Microsoft’s Jay Parikh joined us to discuss how even agent harnesses are getting a major evolution with Microsoft’s newest releases. You.com’s Saahil Jain also sat down with us to explore the big changes we’ve seen in AI tools since 2024, and why heavy orchestration is a thing of the past.

And it’s not just the harnesses and architecture that are changing around AI. We’ve basically added more layers of abstraction on top of our abstractions, which is why we have a piece from the web this week on controlling ideas instead of code. But abstracting everything is gonna cost ya: we’ve got the story of token costs between models, so you can be well-informed when you someday abstract your entire life away.

But not everything in our bits and binary world is feeling earth-rocking shifts on the daily. No, one thing is always going to stay stable: the curiosity of our Stack community, who ask the questions we're all wondering about. For instance…what’s the likelihood of death by space debris? Does Lex Luthor use KryptonGPT? How did this rusty screw get in my laundry? Let’s not shake things up too much and end this Overflow like we always do. We've got all that and more ready for you down below.

From the blog

Building more than just an agent harness

Live from Microsoft Build, Ryan is joined by Jay Parikh, Microsoft’s VP of AI Core, for a conversation on what enterprises need to build, deploy, and run AI agents at scale with demonstrable ROI; how Microsoft built an end-to-end agent development system that goes past just the harness; and how you can evaluate for reliability and correctness in models that get more intelligent and autonomous everyday.

What's left for infrastructure-as-code after AI moves in?

Ryan is joined by Rosemary Wang, Developer Advocate at IBM, to explore what infrastructure-as-code looks like once AI starts writing and deploying it.

Agent orchestration is so two years ago

Ryan welcomes Saahil Jain, CTO of You.com, to discuss why building agents with a 2024 mindset is a mistake as modern models improve at long-horizon tasks, why heavy orchestration layers can hurt model performance more than help it, and why the 2026 competitive edge actually comes from information retrieval and unique data paired with end-to-end evaluation.

When the sensor starts thinking: SnortML, agentic AI, and the evolving architecture of intrusion detection

Signature-based detection has always known what it was looking for. Machine learning and autonomous agents are changing the question entirely, shifting from "does this match a known pattern?" to "does this actually make sense in context?"

Interesting questions

How can I remove a rusted and stripped screw from my dryer cabinet?

Have you tried turning off the screw and turning it back on?

What exactly do we obtain from a concrete formal proof?

Is there any proof that this proof is proof?

Links from around the web

The same TypeScript costs 73% more tokens on Claude than GPT

So is it or is it not cheaper to just get a human to do it?

Control the ideas, not the code

We've abstracted the code, now just to abstract our own minds!

The power of collaboration: How we can reduce traffic congestion

Now to collaborate on using our blinker when changing lanes.


Our sixteenth Annual Developer Survey is now open and we want to hear your thoughts on all things software. Take the survey now!