- Vanagon Briefing
- Posts
- Vanagon invests in 36ZERO Vision
Vanagon invests in 36ZERO Vision
€3.6 million in funding for Tech Wunderkind and Unicorn Founder to fuel zero-defect production with AI
Sandro Stark
July 15, 2025
It’s a big year at Vanagon. As we approach the final close of our Fund I in the coming months, momentum is building fast — we’re welcoming new LPs, expanding our portfolio, and accelerating the growth of the Vanagon DeepDrive platform to support the next generation of technical founders. Today, we’re proud to announce our 17th investment from Fund I: 36ZERO Vision, a company redefining what industrial AI can achieve in the real world.
€3.6M Funding Round to bring AI-powered, zero-defect manufacturing to Europe's industrial backbone and beyond.
I still remember the excitement I felt the first time I saw vision AI leave the labs in Redmond back in 2016 and cameras began to index the real physical world. In the early days, vision AI systems often required millions of labeled images to perform well. But the frontier has shifted. What excites us about 36ZERO Vision is that this technology is no longer experimental. It’s production-ready, with implementation times measured in days and zero churn across their customers. Helping industries reduce waste, improve precision, and raise their quality bar isn’t just good business. It’s the kind of impact we want to see scaled across Europe and beyond.
Several more portfolio companies are still in the pipeline, and we’ll be sharing more soon - stay tuned!
Tech.eu just spoke to 36ZERO Vision’s CEO Heiko Huber and co-founder and CTO Zeeshan Karamat, here is an excerpt from the article:
Built for industry, born in a hackathon
Co-founder and CTO Karamat has a background in mathematics and computer science, with a strong entrepreneurial mindset and work ethic. He shared:
“Even during high school, I was freelancing, working for some of the top 100 companies globally. I was also among the top three students in a national math competition in Pakistan.
I had this extreme learning phase — working up to 22 hours a day using what I called the “Superman pattern”: four hours of work, 20 minutes of sleep, and repeating it five times.
After that, I applied to Stanford and was lucky enough to be accepted, but due to my nationality at the time, I couldn’t get the visa. So instead, I applied to universities in Australia and Germany. In Germany, tuition was free, so it was an easy choice.”
He started working at Microsoft and later at BMW Group, where he led a predictive maintenance project where,“we could detect if a machine would fail within 9–10 weeks, and the accuracy was really high.“ It was a successful project, but he realised the 9-to-5 enterprise life wasn’t for him.
He participated in hackathons on weekends — MIT, Oxford, Microsoft — and ended up winning more than 100 times.
One of the hackathons at BMW Group led to the start of 36ZERO Vision, where his team aimed to do industrial quality inspection purely through software, using minimal hardware.
“Everyone said it was impossible to achieve industrial-grade accuracy on low-end hardware.”
That challenge pushed them to develop their own foundation model. That’s also when he met Heiko.
Huber is from southern Germany, near Stuttgart, and his addiction to cars inspired him to study mechanical engineering in Munich. He co-founded a successful fitness startup in the fitness tech space — the now hugely successful EGYM. He admits, “being a fairly conservative German, I decided to pause that and finish my studies — no regrets there.”
Afterward, he did an MBA and ended up at McKinsey. A role in Siemens’ venture unit, and as Managing Director at UnternehmerTUM, Europe’s largest entrepreneurship centre followed, where he met Zeeshan. He recounts:
“He was part of the program, and I immediately fell in love with the technology and the vision. It was clear to me that this had the potential to bring real efficiency and quality to the industrial world. Munich has the perfect ecosystem for this—talent, customers, infrastructure.
When Zeeshan asked me to join as co-founder and CEO, it was a no-brainer.”
A software-first approach for real-world deployment
According to Zeeshan, the startup originally began using iPhones because they're accessible and easy for customers to deploy without any specialised hardware:
“Our tech stack was developed with low-end hardware in mind. We rewrote the entire foundation model in C++, optimised to run efficiently even on devices like iPhones, NVIDIA Jetson units, or low-spec GPUs.
We support industry protocols like GenICam, which allows us to connect to a wide range of industrial cameras — Matrix Vision, IDS, etc.
On the compute side, we use low-level languages to ensure maximum efficiency.”
A modular, data-efficient foundation model
36ZERO Vision is using deep learning to detect patterns, independent of background noise or lighting. Its model not only learns from real images but can simulate hundreds of thousands of synthetic images based on key variations.
The core of 360Vision’s system is a multi-stage foundation model. Each stage trains on a specific task, like a “mixture of experts.” Each expert learns small, specific things, and collectively they deliver high accuracy.
Currently, the model is trained on 2D visual images, from single-channel to five or six-channel images.
According to Zeehan, “these could be standard RGB images, mono cameras, or even X-rays.”
“The powerful part is that we only need a small number of data points, sometimes as few as five images.
In one benchmark with just 10 images, we achieved nearly perfect accuracy — no missed defects, no false alarms. In contrast, a state-of-the-art competitor had a 9 per cent error rate. Our platform is highly data-efficient.”
A solution to real-world frustration
“We built 36ZERO Vision to solve a real-world frustration," said CEO Heiko Huber.
Visual inspection too often only works in theory, but fails in production. Most systems still rely on pixel-perfect comparisons, which means that even small differences, reflections, shadows, and angle shifts can cause false positives.
Most of 36ZERO Vision’s customers already use some form of computer vision, but according to Huber, “they’re frustrated. It takes hundreds or thousands of images, plus manual labelling. Even then, they often get too many false positives.
According to Huber, "our system delivers higher accuracy with lower complexity, and it’s already earning the trust of top-tier manufacturers.
"This funding helps us scale faster and push the limits of what AI in manufacturing can do.”
The company’s proprietary AI technology is already industrialised and actively deployed in production environments at some of the world’s most demanding manufacturing companies, including Siemens, Bosch Rexroth and LEONI.
Further, 36ZERO Vision’s technology significantly outperforms established competitors such as Keyence and Cognex by delivering higher accuracy, reducing inspection errors, and simplifying the inspection process.
Significantly, with 36ZERO Vision’s solution, users only need a handful of images. Even more importantly, they can label them in their cloud-based platform, train the model, and validate performance themselves. It’s a self-serve onboarding, with no hardware installation or on-site integration required.
I was curious how open manufacturers are to switching from existing solutions. According to Huber, “they’re surprisingly open—because the pain is real.”
“That said, it’s a conservative industry. One-off licenses happen fast, but full rollouts take time. They want to be sure the system works. But once you're in, you're in. We’ve had almost zero churn.”
In some cases, 36ZERO Vision has fully automated inspections that were previously manual. This kind of deployment can achieve ROI within 4–5 months — that’s exceptional in industrialtech.
36ZERO Vision started in automotive, then moved into electronics, machinery, and building materials. It has growing interest from the defence sector, where the product is a perfect fit for its high quality requirements and complex inspection needs.
According to Huber,
“Eventually, we’ll move into semiconductors — wafer and chip inspection — but that’s likely two to three years out.”
This latest investment round was led by an investor syndicate including Join Capital, Bayern Kapital and Vanagon Ventures, with additional support from notable investors including UnternehmerTUM Funding for Innovators and Alchemist.
Manufacturing is overdue for AI-driven innovation that reliably delivers results beyond the lab. 36ZERO Vision’s breakthrough solution is rewriting industry standards by dramatically cutting false positives and simplifying processes.
Building the future of industrial AI in Southern Germany
With the new funding, 36ZERO Vision is focused on scaling across Europe, expanding with existing customers, hiring, and building more features.
According to Huber, “We’ll stay focused on Europe for the next two years. Long-term, we believe we can build a category-defining company from southern Germany.”
"Ultimately, what we’re building can’t be done in a vacuum. You need proximity to real industrial use cases. AI alone isn’t enough; you need something that works reliably on the shop floor.
Southern Germany has the perfect ecosystem for this: world-class talent, customers, and industrial champions who are willing to adopt new tech. That’s our home base.”
The new capital will be used to expand 36ZERO Vision's team across sales, customer success and product development in response to rapidly increasing customer demand. It will also enable the company to enhance its technological leadership further and reinforce its unique position in the global market for AI-driven visual inspection solutions.