Software ate the world, as the saying goes. But someone still needs to build the fork. And the sensor. And the robot arm that picks up the fork when you drop it. Atech, a Swedish startup that launched today, thinks the gap between 'I have an idea for a device' and 'I'm holding a working prototype' should be about five minutes. Not five months.

The company just closed a pre-seed round with backing from Nordic Makers, Emblem, Lovable (the AI software builder it's often compared to), and Sequoia Scout Fund. The round size hasn't been disclosed, but the investor names say plenty about where the smart money thinks physical AI is headed.

Here's the pitch: you describe a hardware concept in plain language. Atech's platform generates the design, selects from a library of modular electronic components, writes the firmware, and produces something you can actually hold. If Lovable lets non-engineers ship software products, Atech wants to do the same for the physical world.

The team behind Atech isn't coming out of a weekend hackathon. Vladimir Baran, Tomas Erik Harmer, and David Stalmarck bring backgrounds spanning embedded systems, industrial design, and hardware manufacturing. They've watched the Maker movement, the Arduino ecosystem, and Raspberry Pi era promise democratized hardware creation. Each wave made things cheaper. None of them made things truly easy.

The bottleneck was always the same: you needed to know electronics, firmware, mechanical design, and manufacturing constraints simultaneously. That's a rare skill stack. Most people who have great physical product ideas don't have it. Most software engineers certainly don't.

Atech's bet is that large language models and generative AI have finally crossed the threshold where a machine can bridge that knowledge gap. Not perfectly. Not for every use case. But well enough to get from zero to a functional prototype that a human can then refine.

The platform works with a component library: OLED displays, environmental sensors, LED grids, motor drivers, touch sensors, joysticks, haptic motors. Each module has standardized interfaces, I2C, analog, digital, that let them connect without custom wiring. Think of it as a vocabulary of hardware parts, and the AI as the grammar that assembles them into something coherent.

You describe what you want. The system selects modules, generates connection logic, writes the firmware (likely MicroPython or C++ for ESP32-class microcontrollers), and presents you with a build guide. The ambition is that a working prototype arrives at your door shortly after, assembled from the platform's modular parts.

The timing isn't accidental. Physical AI, the convergence of large models with real-world hardware, is becoming one of the hottest investment themes of 2026. Nvidia's entire roadmap bends toward it. Robotics startups are raising at valuations that would've seemed absurd two years ago. The demand for people who can build hardware that talks to AI systems vastly outstrips the supply.

Atech's thesis is that this talent shortage is itself the opportunity. If you can't hire enough hardware engineers, let the machine do what the machine can do, and let humans focus on the creative decisions. Which sensor goes where. What the product should feel like. How it should respond to its environment.

It's a compelling pitch. But hardware startups carry risks that software startups don't. Supply chains break. Components go out of stock. Physical products need testing that can't be simulated away. The 'Lovable for hardware' analogy is attractive, and it's also a high bar. Lovable works because deploying software is essentially free. Deploying hardware never is.

One name on the cap table deserves extra attention. Sequoia Scout Fund doesn't write large checks, that's not the point of the scout program. But Scout investments have historically been early signals of where Sequoia's broader interest is heading. The fund backed Stripe at inception. It backed Notion before anyone else.

A Scout bet on a Swedish physical AI startup suggests that the firm sees hardware democratization as a genuine category, not a niche experiment. Combine that with Lovable investing in a company explicitly modeled on itself (for a different domain), and you get a pattern that's hard to ignore.

Atech is pre-product, pre-revenue, and pre-everything except ambition. That's fine. At the pre-seed stage, you're betting on the founders and the thesis. The thesis here, that AI can make hardware creation as accessible as software development, is either wildly optimistic or exactly right. The next twelve months will tell us which.