With its robust and agile backend, MindSphere, Siemens industrial Internet of Things as a service solution, is only as strong as its partnerships. To help power the amazing artificial intelligence capabilities found in MindSphere, Siemens turned to our partner Tangent Works.
Tangent Works was founded in 2014 by a team of data scientists and mathematicians focusing on predictive modeling for optimizing operations and reducing costs. Headquartered in Belgium, the team created the Tangent Information Modeler, or TIM, an automatic predictive model building engine that uniquely automates the forecasting and anomaly detection process by analyzing time series data and generating accurate models based on the patterns it detects. We spoke with global partnerships director Sam Verdonck about the partnership with Siemens, the AI capabilities found in MindSphere and the future of Tangent Works and Siemens.
Q. How long has the Tangent Works been partnering with Siemens?
A. We officially signed as partners October of last year. But we started this engagement with the Siemens Rocket Club in 2019, which is an AI competition of Siemens, hosted by the AI lab in Munich. Then, in 2020, we went into business development mode and testing and validation. For the last one and half years, we we’ve been building applications and solutions together.
Q. What is Tangent Works involvement with MindSphere?
We are a tooling company. Any software company works with a stack of tools they use to build any software or application. We are now deployed at MindSphere on the MindSphere AWS environment and the Microsoft Azure environment, meaning that all MindSphere product management has direct access to TIM to build their applications.
At the moment, the focus is to build the AI for Everyone application, as part of Operations Insight in MindSphere. First, to develop an app for essential forecasting and an app for essential anomaly detection. But there are other apps in the pipeline, including advanced forecasting, advanced anomaly detection, and root cause analysis. And also more use-case-specific applications like energy optimization or product quality monitoring – these all use our technology under the hood but the customer doesn’t even know it.
Further on the roadmap is the launch of “what if analysis” capabilities to further support digital twin applications. With TIM we can feed the comprehensive digital twin model and explore and optimize unlimited “what if” scenarios, providing value not only to the field but the entire enterprise.
The current apps are focused on exploration and experimenting. We’re giving the power of AI and machine learning to all users, not only to data scientists and that can help to improve data literacy. A lot of companies have ambition to become more data driven, but people who really work with the data are only one small corner of the company. We’re giving a much bigger audience the ability to test and experiment and explore with data, now that it has the strength of MindSphere cloud enabling. In a couple of seconds you can have a model, an experiment. If you don’t like it, you press a button and you have a new predictive machine learning model. This takes a data scientist hours or days.
Q. Can you expand on how AI for Everyone makes AI more accessible?
By automating the complexity. The most complex parts get automated, so you can do everything with one click of a button. Of course, you still have to decide, for example, how far ahead you want to forecast. If, for example, want to forecast the next 24 hours or 1,000 runs. But this lets an engineer that’s not a data scientist, really start using it.
That’s partly where other techniques by other vendors and other tools come short. They look at what the data scientist has to do and they bring in brute compute. This consumes a lot of computing power. We position ourselves as green machine learning. We consume less power to build a model than auto ML. It’s much leaner, more efficient and greener.
Also, time series model usually gets obsolete very quickly, but with the automation we offer, it’s always up to date as the time series data evolves catering to reliable large-scale production scenarios.
Q. What are the benefits of partnering with Siemens and MindSphere?
We are a product company and we have a generic time series machine learning toolkit, but we are a bit far away from use cases. They usually come through partners. Our product is an engine that builds models. But if you want to visualize that or you want to use that data to trigger events or rules or work orders or alarms and so on, you need platform integration – and ultimately solutions and applications that address specific customer use cases.
With MindSphere, and also Mendix, you can create databases and management tools for IoT data. Once it’s there, you can start leveraging for insights. That’s where the business value of TIM really gets delivered for specific use cases. This is a very promising partnership, with a lot of potential.
Q. What is next for Tangent Works and MindSphere?
There are several things in the pipeline. Today it’s mainly about the exploration and experimentation by all the different profiles. So I have this piece of data. Let’s see if there’s predictive value. Let’s very quickly and easily do a test experiment. But very soon, applications will be launched to schedule predictions. Let’s say you create a forecast or an anomaly detection. You like it. Soon, you can schedule it do that forecast or detection every time new data is available.
We’re also working on root cause analysis. That’s very exciting, because if you detect an anomaly, you want to know why there is an anomaly. Right now, the manual task of analyzing it becomes something that happens in a corner of the company with people who find extra time to dive deep on the machines’ failing. But our product helps to do this more automatically, so engineers and manufacturers can optimize their business processes in a more integrated and pragmatic way, and not in silos.
Q. What makes your partnership with Siemens unique?
We are also a Microsoft and AWS partner, so, our partnership with Siemens is not exclusive. The difference with the Siemens partnership is that it goes much further. With MindSphere, more out of the box applications are being built by Siemens itself with our support, because Siemens has the domain knowledge around industrial IoT.
And also with Mendix app development, customers have the opportunity to build their own low-code applications with powerful time series machine learning. With other partners, it’s still much more dependent on coding and third-party consultancies.
Interested in learning more about AI and the industrial Internet of Things? Check out our FAQ “AI with IoT made simple: MindSphere AI for Everyone.” Then, once you’re ready to explore the capabilities of MindSphere, start for free.