Giving future robots a digital brain: Startup cuts years off AI development

22 October 2026

The startup Synteza has cracked the code on generating computer-created training data in record time. This is set to make it significantly faster and cheaper for robots and autonomous systems to understand the physical world. The company is now moving into the robot incubator at the Danish Technological Institute, backed by the Odense Robotics StartUp Fund.

Autonomous machines, drones, and humanoid robots are expected to become an increasingly prominent part of our society and workforce in the near future. Common to all of them is that they are packed with cameras and sensors to perceive their surroundings. But before these robots can operate safely and effectively in practice, they must learn to interpret and respond to what they see.

And that requires massive, almost overwhelming amounts of training data.

The startup company Synteza aims to solve that challenge. The company has just been admitted to the incubator at the Odense Robotics StartUp Fund at Danish Technological Institute (DTI).

– We founded the company because we saw a huge need for specific data and the right infrastructure to make the things around us autonomous. The single biggest problem for artificial intelligence is finding the right data so that systems can function in the physical world. We solve this by creating these data sets digitally and training physical AI systems directly within our digital pipeline, explains Serhii Lubkin, CEO and co-founder of Synteza.

From years to weeks: Putting an end to endless data collection

Traditionally, working with data is a heavy and time-consuming bottleneck when developing AI. First, vast amounts of raw data must be gathered from the real world, after which they must be sorted, cleaned, manually labeled, tested, and validated. This is an iterative process that often takes months or years.

Synteza turns this approach on its head. Instead of gathering data out in the physical world, they build and model data sets digitally.

– We all expect humanoid robots to be everywhere around us soon. They look at the world through cameras and sensors, but they need to understand what is happening. Teaching a robot what even a toddler grasps intuitively usually requires immense amounts of data. Instead of hauling around these massive volumes of collected data, we create the data digitally. We do not need to collect vast amounts from reality; we can produce them tailored to specific needs, explains Serhii Lubkin, elaborating:

– This means that a process that previously took years can now be completed in weeks or even days, with the exact same high quality. It fundamentally changes how models are trained, making it far easier and faster to bring humanoid robots and other autonomous systems into the real world.

The technology is also relevant for manufacturing right now, where it can be applied to automated quality control on factory floors:

– When manufacturing components, you need to catch defects and flaws down to the smallest detail. But to train an AI to do that, you need data, and most often, there are far too few examples because defects fortunately do not occur all the time. We have developed a method to generate precisely the data sets that are missing, enabling AI models to quickly learn how to spot production flaws, says Serhii Lubkin.
 

Serhii Lubkin, CEO and co-founder of Synteza

United by a shared vision

Behind Synteza are Serhii Lubkin and his co-founder, Oleksandr Pak. Serhii has a background in AI and computer vision, where the lack of available data, according to him, has always been the single greatest hurdle to creating models that genuinely perform in the real world.

– When I met Oleksandr, he presented the idea of generating training data digitally with unprecedented precision. I immediately realized this was exactly what the industry needed to make physical AI a reality, and we decided to join forces and build the company, Serhii Lubkin explains.

A unique robotics ecosystem in Odense

Synteza is now moving into the incubator at DTI in Odense. Through the Odense Robotics StartUp Fund, the company gains access to founder-friendly capital, a strong mentoring network, and testing facilities in the robotics halls.

– We applied to the Odense Robotics StartUp Fund because we truly value the ecosystem here. It brings together so many facets of physical AI, including robots, drones, and autonomous vehicles. It is an exceptionally strong environment with solid connections both in Denmark and internationally, says Serhii Lubkin, concluding:

– Our primary goal in the incubator is to forge close ties with the Danish robotics ecosystem and launch pilot projects in collaboration with DTI. It is a fantastic environment to be part of, and we look forward to both benefiting from it and contributing to its continued growth.