Bosch Power Tools' metamorphosis into a data company with Kafka

'The overarching goal is to connect all the devices so we can have an overview from birth to trash'

Every year, Bosch Power Tools, a division of German industrial giant Bosch, churns out hundreds of thousands of drills, mowers, trimmers and power saws for DIYers, as well as more heavy-duty equipment used on construction sites. It's very much a hardware business, but over the years the data produced by and about these tools has become almost as important as the devices themselves.

There's data stored in CRM and sales platforms about who bought and registered the tools; there's sensor data about how the tools are used in the field; there's information from repair shops about where and how tools have failed; and there's data flowing from the production lines. The aim of the Power Tools IT team is to take all this data and combine it in novel ways to add value for customers and the company alike, said solution architect Ralph Debusmann.

"The overarching goal is to connect all the devices so that we can have an overview of all tools from birth to trash," he told Computing.

Debusmann and his colleagues have spent the past four years building a platform that's capable of pulling together and interrogating all this data in a way that meets the company's long-term business requirements. Already in production are: a DIY and Garden app to provide relevant offers for upgrades and help with DIY projects; tool runtime optimisation - collecting behaviour-based field data and using AI to feed back improvements, such as how to create the perfect lawn; enhanced product registration procedures; and RefineMySite, a JIRA-like platform for managing building sites and tool deployment.

The team is also rolling out predictive maintenance, using field data and aggregated intelligence to warn users to have parts replaced before they fail. Available now for a few tools it will be rolled out to a wider range soon. There will be much more of these "AIoT" (AI+IoT) applications in the future, Debusmann said.

Many of the applications, both current and planned, have a strong real-time element. Examples include detecting fraud during orders and registration, and providing immediate feedback to tool users in the field. For this reason they chose Apache Kafka as the backbone for the platform, rather than a batch-oriented data lake approach. Distributed stream processing platform Kafka can store large amounts of data as well as providing for rapid and ordered throughput of data streams. These capabilities provide the low latency and resilience that Bosch Power Tools' microservices-based IoT platform needs, Debusmann explained.

"Our apps talk to the backend which is running on Kubernetes. We completely rely on Kafka for communication between microsoervices and with the outside world. The advantage is, when you connect microservices to each other they do not rely on the other microservice to be online, because you have Kafka in the background which is pretty much guaranteed to be online all the time. So, there's a stable backbone behind the data flows, and even if one of the microservices falls down you can restart it and you don't lose any data."

Because operational aspects of Kafka were not a core competence, they opted for a managed Kafka option with Confluent Cloud.

"Bosch Power Tools is not traditionally a software house. We're getting bigger and bigger, but we started out with a pretty small team, and we were not able to do our own Kafka hosting. So we needed a partner who takes care of Kafka completely and that's what Confluent does, really it's like plug'n'play Kafka."

The future will see many more devices connecting to the cloud, Debusmann says, with existing devices such as battery chargers fitted with sensors and Wi-Fi transmitters and streaming data back home. A virtuous circle, this will allow hugely more powerful analytics from combining these streams with data from manufacturing, repairs and the field to feed into the design and manufacturing process, and with CRM and sales data to enhance marketing and direct sales.

The Bosch Power Tools IT team is working on providing bridges between batch-based analytics tools and visualisation solutions such as Power BI and Tableau so that dashboards can be built for fraud detection, predictive analytics, marketing and other use cases.

It's all part of Bosch's steady metamorphosis into a data company, where information, AI and advanced analytics are seen as fundamental to staying ahead.

"AIoT really means to enrich IoT with more AI, since this is where Bosch as a whole wants to become big in," Debusmann said.