There’s a shortage coming that nobody’s put on a slide yet: senior engineers. Every company on earth is about to want one who can drive a fleet of agents with judgment — and almost nobody is making new ones anymore, because AI ate the entry-level work juniors used to learn on. You cannot conjure a senior in a quarter. We just watched this exact movie with memory: DRAM prices went vertical because demand exploded and you can’t build a fab overnight. Senior engineers are next, and the only hedge is to build your own, starting now. The best thing you can build in the agentic era isn’t your product — it’s another Engineer. I have juniors on my team. This is my problem, right now. Here’s how I’m thinking about it.
Here’s a question worth more than most of the AI hype takes clogging your feed: what happens to your velocity the day your frontier provider changes the deal? Raises the price. Deprecates the model you tuned your whole workflow around. Throttles you at the worst possible moment. Or just decides your use case, your industry, or your country isn’t one they want to serve anymore. If your honest answer is “I’d be dead in the water,” then you don’t have a strategy — you have a dependency. And it’s time to look hard at running some inference on your own metal.
Here’s a result that surprised me: the reasoning model wrote more correct code than the plain one — and still lost. It nailed all 32 test cases, then sat there and refused to declare itself done. Burned the whole time budget re-checking work that was already right. “Correct, but won’t stop” turns out to be a real failure mode, and it’s exactly the wrong one for an autonomous agent loop.
XDA ran a piece that stopped me mid-scroll: a 27B open-weights model reverse-engineered a commercial application’s license check — recovered a deliberately obscured crypto key out of ARM64 assembly, caught and corrected its own mistake without being told, and produced a working bypass PoC. In about thirty minutes. On a desktop box. I have (a version of) that box. So I went and set it up.
I have two 128 GB unified-memory machines on my desk. Day job gear, not mine. One is an ASUS Ascent GX10 — an NVIDIA GB10 Grace Blackwell appliance half the size of a hardback book. The other is an HP ZBook Ultra G1a, a 14" mobile workstation running AMD’s Ryzen AI MAX+ PRO 395 (“Strix Halo”). Both hold an 80B coding model entirely in memory. How do they compare for agentic coding?
Right now, as an industry, we are running almost entirely on COGS and calling it innovation. Kelsey Hightower gave a talk at PlatformCon called Zero Token Architecture, and the whole thing compresses down to one sentence: infer once, export, run without inference. He’s right. I want to give you the vocabulary that makes it land in a budget meeting, because I work at a company that ships hardware, and we already have words for this.
Somewhere in the last year, a bunch of us who hadn’t shipped real code in years quietly started shipping again. Not reviewing it. Not architecting it in a doc and handing it off. Actually building it. And loving it.
Python won because it was easy. That was the whole trick. And now, in the age of agentic coding, that trick has stopped being an advantage — and quietly became a liability.
Last post I showed off the chainsaw I built myself — my build-autonomous loop, the padded room, the whole rig — and cut down a little tree with it: a Go CLI to boss around a smart plug. It was well worth building! I also think you should probably stop building your own. Yes, including me. Especially me.
People keep asking me how I actually work with Claude Code now. Not the “does AI coding work” question — I’ve beaten that horse into glue on this blog already — but the boring, practical, how-does-my-loop-actually-work question. So here it is. The whole rig. And fair warning: what I’m about to describe is a chainsaw, and most of the industry is still lined up at the axe-chopping contest.
Look, I’m going to be direct here. If you are not using AI agentic coding methods right now - today - you are going too slow. Not “a little behind.” Not “maybe missing something.” Too damn slow.
Open source is about to get a lot messier. Not because the code is getting worse – because AI agents just changed the economics of “fork it and fix it yourself.”
Software development is stratifying into three distinct tiers. Where you land will determine whether you make more money, the same, or far less than today.
In my previous post about Claude Code, I talked about using VS Code devcontainers to safely run Claude in “dangerous mode.” That was great for a typical software project. But what about something harder? What about porting a complex hardware-interfacing library from Python to Go, where you need to deeply understand USB protocols, radio registers, and firmware internals?
That’s exactly what I did with gocat – a Go library for controlling the YardStick One sub-GHz RF transceiver. And Claude Code was instrumental in making it happen.
So I’ve been working on this project called client-w-mcp – it’s a leanring project to truly understand how an AI agent works with MCP servers. And I’m exploring Agentic development - with Claude Code.
Why Claude Code?
The first time I used it, Claude just… flowed. It seems to do a lot more by itself to figure things out. I especially like the Task() so that it can go do more than one thing at a time. It would be bad to try to modify code that way, but to write tests or update docs and not have to wait doing one thing at a time… is wonderful.