Juna.ai GmbH, a German startup using artificial intelligence to make factories more efficient, launched today with $7.5 million in seed funding.

Kleiner Perkins led the investment. It was joined by Stockholm-based fund Norrsken VC, prominent venture capitalist John Doerr and other backers.

Juna.ai is led by Chief Executive Matthias Auf der Mauer, whose previous industrial software startup was acquired by publicly traded sensor maker Sensirion Holding AG. Co-founder Christian Hardenberg, the company’s chief technology officer, previously held the same role at Delivery Hero SE. Juna.ai provides a hosted AI platform that promises to help manufacturers find ways of operating their production lines more efficiently.

The company’s platform integrates with applications from SAP SE, Snowflake Inc. and other enterprise software providers. It uses those integrations to collect technical data such as equipment temperature readings from customers’ plants. From there, Juna.ai’s AI algorithms turn the raw data into dashboards that highlight areas for improvement in a factory’s operations.

Some of the platform’s features focus on helping manufacturers boost production reliability. Juna.ai can measure the number of products that a factory produces per day, the frequency at which manufacturing defects emerge and related metrics. The software also generates improvement recommendations. 

Reducing factories’ energy usage is another task to which customers can apply Juna.ai’s platform. For manufacturers that produce multiple types of goods, the software breaks down electricity usage by product. Juna.ai also identifies situations where a production line consumes more power than usual and provides technical data that can be used for troubleshooting.

According to the company, customers can use the information produced by its platform to bring their plants into compliance with the ISO 50001 standard. This is a collection of best practices designed to help manufacturers cut their electricity usage. Additionally, Juna.ai turns the power consumption data it collects into reports that can be used to demonstrate compliance with manufacturing regulations. 

Under the hood, the platform is powered by custom AI models. The company says that it developed its models using a method known as reinforcement learning. Neural networks based on this technology are trained through trial and error: They perform a task repeatedly to learn the best way of completing it and receive positive feedback after each learning milestone.

Juna.ai fine-tunes its models on customers’ data to produce AI agents. Those are customized neural networks that can perform tasks with little to no human input. According to Juna.ai, each of its agents is designed to monitor and optimize a different set of industrial metrics.

The company currently has six employees. Juna.ai will use the newly announced seed round to hire more staffers and grow its customer base. 

Photo: Unsplash