Danish Atech is amongst a small handful of new AI startups that have launched AI products to make hardware prototyping much, much faster. Learn about their story and how AI may completely change the game of creating hardware
“Guys, it’s 2026. AI has completely revolutionized software, yet hardware has barely changed since 2005. At this point the gap is getting ridiculous.”
Tomas Erik Harmer is speaking to a crowd of hardware startups, investors, and manufacturers at an event hosted by the accelerator program Future Manufacturers, which helps startups go from prototype to production. He is convinced AI can change the game for hardware the same way it has for software.
“Everyone will soon be able to build hardware,“ he continues.
His own company, Atech, and a small number of other international startups are currently building AI solutions for hardware prototyping. Atech is only five-month-old and has recently launched its first version of an AI platform and physical “building blocks” to create fast prototypes.
How it works
Users buy a starter kit of modules and a motherboard, then describe their product idea to Atech’s AI. The platform generates the hardware configuration and firmware for a prototype using those modules.
From watering plants to prompting hardware
The idea for creating Atech sprang from a whole different startup idea. Harmer and his former cofounder Vladimir Baran both wanted a smart device to help them keep their plants alive – a plant pot watering the plants when needed. This couldn’t be hard to prototype. Or so they thought:
“It seemed like a very easy product to make because it was very simple, but we were just shocked at the complexity of building a hardware product. I think it took seven iterations of like two to three weeks or even a month each to get to a working product.”
This sparked a whole different idea for a startup, and Atech was born.
“I really liked Lego and Minecraft growing up, and I kind of got a bit of inspiration from those to say like ‘Hey what if we just turned hardware into these modules and then combined it with AI?’,” Harmer says.
Atech’s building blocks are custom-made modules like sensors and cameras, which can easily be connected to a motherboard. This eliminates the need for wiring and the need to understand in detail how each component works.
A huge potential
Atech has already reached $800.000 in funding – one of the investors being Lovable, which has been the company’s biggest inspiration, according to Harmer. The Swedish company Lovable is a huge “vibe-coding” success that helps users prompt websites and apps and was recently valued at $13.3 billion.
The market for vibe coding products is booming, and the potential is huge in hardware as well.
The most direct competitor to Atech is – according to Harmer – the startup Schematik. Schematik differs a bit from Atech as it works across various hardware and demands more from the user providing wiring instructions instead of using predefined modules. This makes it more flexible but also harder to work with. Atech is currently working on opening their hardware to off-the-shelf components but aims to keep the system simple to use for non-engineers. In its year of foundation – 2026 – Schematik managed to raise $4.6 million.
Another player is Diode, which was founded in 2024 and focuses solely on PCB design through prompts. It brought its total funding to $14 million in 2025. So if you “follow the money” – these startups definitely have ideas worth noticing.
Atech
Atech is a Danish startup founded in 2026 by Tomas Harmer and David Stålmarck
Atech has developed an AI-solution for prototyping hardware.
The company has raised $800,000 in funding.
Read more: https://atech.dev/
Game changer
And those in the hardware business need to pay attention.
“It will save a lot of time, and it is even more meaningful to adopt hardware solutions. It will accelerate production more – especially as it’s always more difficult to find engineers designing hardware than engineers doing software,” Assistant Professor at DTU Charalampos Orfanidis says.
Orfanidis specializes in embedded systems engineering, with a specific focus on merging hardware constraints with artificial intelligence – and is also a former advisor to the founders of Atech.
He believes the current AI solutions for hardware are “very promising”.
“It may be a little too early to say: “this is a breakthrough”, but we have some first examples and know that we can produce hardware with AI. That is amazing. I’ve also seen Atech’s demo, and it’s impressive, and it seems to work.”
Orfanidis is certain that, once the technology reaches wider adoption, it will be a game changer, just as AI has been in software development.
The risks and trade-offs
So, we know it is possible. And we know the solutions are coming.
However, there are some risks when using AI in hardware. If you ordered 1000 units and only later realize some small AI-hallucinated mistake, it will be expensive. With software, you can easily just go back and edit some code.
And there is the case of debugging if the hardware prototype doesn’t work.
“To debug there, you will have to do it in real life, and you would need a multimeter and put a probe on the circuit and measure the voltage or the current to see if it’s working or not. So, it’s way more difficult, the whole process in general,” Orfanidis explains.
This is not the only pitfall. John Teel is an electronic product-development adviser and former Texas Instruments microchip design engineer who spent more than a decade designing integrated circuits before founding Predictable Designs and the Hardware Academy.
He has more than 88,000 subscribers to his YouTube channel, and in a video, he explains that there are procedures, physical tests, and standards that AI in general does not take into account. Among other things, he highlights the following questions when choosing components:
- Will the manufacturer still make it in 5 years?
- Can you buy 100,000 units next quarter?
- Is there a pre-certified version?
Another potential issue is whether or not we will benefit from AI designing our hardware in the long run. The more AI takes over, the more our product design becomes a black box.
“Some software companies that did layoffs now hire people back. They need them to debug or document code generated from AI. Because maybe they have products that are working, but they don’t really understand how this works,” says Orfanidis.
The dream
Ultimately, the Atech team hopes to make production easy for any startup – or for that matter any company or person. And their product is already out there to make that possible.
“We’ve sold thousands of modules. A lot of our customers are in Sweden and in the United States,” Harmer says.
By October, they plan to launch a version bridging the gap to external, off-the-shelf hardware. This is enabled by so-called “smart pins” on the motherboard designed to safely connect with third-party components.
To prevent breakdowns when adding these new components, Atech has added a monitor to the circuit board that reports what’s happening in the physical world.
“The idea, though, is that it will protect itself. So, if something’s going wrong, it’ll shut itself down, and you don’t need to be involved in any way,” Harmer explains.
In the end, the Atech team wants to prompt finished products that can go straight into production.
“We’re going to automate this in the future so you can just, like, one-click to mass order a cheaper version of your product with the same functionality,” Harmer explains.
Future Manufacturers
Future Manufacturers is a community and a startup program made in a collaboration between Danish Industry and MADE with the purpose of helping hardware startups go from prototype to full-scale manufacturing.
