I read this Ars Technica piece about a chap who bought a Unitree robot dog for $4,000 and put it through its paces, and it's stuck with me for a few days now. Not because I'm about to buy one — though I'll admit the temptation is there — but because of what it quietly tells us about the next phase of automation.
I've spent nearly forty years watching computing get cheaper, smaller, and more capable at a pace that still catches me off guard. But there's a pattern worth noticing: the interesting shifts rarely happen when the technology is new. They happen when the technology becomes boring enough to be affordable. That's what's happened with legged robotics, and it's happened faster than most people clocked.
From Novelty to Commodity
A robot dog for $4,000 that can climb stairs, carry a payload, and run for hours isn't a toy anymore — it's a commodity industrial component. Ten years ago this was Boston Dynamics territory: eye-watering budgets, research labs, defence contracts. Now it's a credit card purchase from a company that also makes affordable robot arms and humanoids. Unitree isn't selling wonder. They're selling infrastructure.
That's the same trajectory I watched happen with content automation. GPT-3 was a research curiosity behind a waitlist. Eighteen months later, everyone building tools like mine was wiring it into pipelines that generate articles, videos, and marketing copy at a scale that would have needed a small agency a few years earlier. The pattern is always: expensive and impressive, then cheap and ubiquitous, then invisible — just part of how things get built.
Robotics is entering the "cheap and ubiquitous" phase right now, and software people should be paying attention, because that's usually our cue.
The Bit Nobody's Automating Yet
Here's what strikes me most: these robots are extraordinary at the physical layer — locomotion, balance, obstacle handling — but they're still fairly dumb about what to do and why. The intelligence bolted on top, the decision-making, the workflow logic, the "go check the warehouse shelf and flag anything out of stock" layer, is thin or entirely absent. It's the same gap I saw in content automation before large language models matured: the mechanics were fine, but nobody had built the layer that turns raw capability into a repeatable, sellable outcome.
That's the opportunity. Not building better robot dogs — Unitree and others are already racing hard on hardware — but building the orchestration layer that sits above them. Task planning, monitoring, reporting, escalation, integration into existing business software. That's a software problem, not a hardware one, and it's exactly the kind of problem I've spent my career solving in the content and marketing space.
What a "Masher for Robots" Might Actually Look Like
I'm not pivoting Safe Online Marketing into robotics — my workshop is Worcestershire, not Shenzhen. But it's worth thinking through the shape of the tool that would matter here, because the principle is identical to what RSSMasher or Article2Video do: take a messy, high-volume input, apply consistent automated judgement, and output something a human doesn't have to babysit.
For robots, that input isn't RSS feeds or video transcripts — it's sensor data, camera feeds, task queues. The output isn't an article, it's a completed physical task with a log and an alert if something needs a human. The transmutation is the same idea: raw material in, useful, reliable gold out. Someone is going to build the "content automation for physical space" platform, wiring cheap robot hardware into ordinary business workflows the way we wired cheap LLM APIs into content workflows. It won't be the robotics companies who win that layer — hardware makers rarely do software orchestration well, historically. It'll be the software people who show up once the hardware's boring enough to trust.
The Real Lesson
The headline here isn't "robot dog is cheap now." It's that physical automation just crossed the same affordability threshold that content automation crossed a few years back, and the winners in the next wave won't necessarily be the ones who build the best hardware. They'll be the ones who build the thinnest, smartest software layer on top of hardware that's already good enough.
I've built my business on being early to that realisation in one domain. I'd be watching very closely if I were building in another.
— Wayne