Tangent Works Puts Machine Learning Modeling into Business Users’ Hands Using AWS

Tangent Works helps companies get more value from their time-series data by automating the machine learning modeling process. Its Tangent Information Modeler (TIM) technology makes artificial intelligence technology accessible and affordable to businesses that don’t have dedicated data science teams. Tangent Works used AWS to launch its services, manage customers’ compute and storage needs, and support its rapid growth.

Many companies struggle to realize benefits from the information they hold about their operations and customers. The shortage of data scientists, who have the skills to analyze data to get useful insights, makes this problem even more difficult to solve.

Belgium-based Tangent Works provides businesses with a fast, affordable way to derive value from their data. Its technology helps customers automate the machine learning modeling process so they can easily perform complex analysis to drive smarter decision-making.

As a young company with a small IT team, Tangent Works turned to Amazon Web Services (AWS) to provide an efficient way to launch its services, manage clients’ compute and storage needs, and support its rapid growth. Using AWS, staff are able to focus on product development rather than infrastructure maintenance, and Tangent Works can dynamically and cost-effectively scale resources to meet variable customer demand.

The beauty of Tangent Works’ tools is that they’re easy to use. Not only do they put powerful machine learning technology into the hands of business users, but they also help data scientists improve their productivity by reducing repetitive modeling tasks. 

The company’s Tangent Information Modeler (TIM) technology—which provides customers with bespoke artificial intelligence (AI) capabilities—delivers the accuracy of manual modeling in a fraction of the time. This means organizations save money both on the staff resources required to create the models and the compute resources needed to run them. “Using AWS, we can instantly scale compute capacity when clients build new models or apply existing models,” says Elke Van Santvliet, machine learning expert at Tangent Works. “Our customers can create models in a few seconds or minutes—this would have taken weeks before. And it’s cost-effective because we pay only for the resources we use.” 

To manage customer workloads, Tangent Works uses Amazon Elastic Kubernetes Service (Amazon EKS), a fully managed container service for Kubernetes applications, and AWS Fargate, a serverless, pay-as-you-go compute engine that automates server management. It also uses Amazon RDS for PostgreSQL, which makes it easy for Tangent Works staff to set up, operate, and scale PostgreSQL deployments in the cloud.

Tangent Works helps customers across a wide range of sectors use TIM to improve their operations. For example, retailers more accurately forecast consumer demand based on historical sales data combined with weather forecasts. Utility companies plan their maintenance schedules taking into account seasonal changes. Energy providers predict consumer usage to keep equipment running at peak efficiency. And financial services firms employ TIM’s anomaly detection to automate credit card fraud detection and rapidly build models in response to new threat types.

Because models are easy to update and run, customers can adjust them regularly, to keep business insights fresh. For instance, retailers can amend the predicted performance and stock requirements of each shop every week, or manufacturers can instantly build a new model when a shop-floor process changes.

Tangent Works is a fast-growing firm that’s always adding customers. And because many of these businesses already use AWS, it simplifies onboarding and collaboration. “Getting our clients up and running quickly means we’re able to grow rapidly,” says Van Santvliet. The company can easily scale its resources to run these increased workloads. Offering its services directly to customers on the AWS Marketplace has also supported growth. Tangent Works has shortened its sales cycles and won 5 enterprise customers in the last 2 months through the marketplace.

The large quantities of data generated by Internet of Things (IoT) systems makes this an ideal use case for Tangent Works technology. TIM provides specialized capabilities for businesses using IoT, including sensor monitoring and anomaly detection. The system is also capable of analyzing failures to improve its predictive abilities.

Tangent Works partnered with Siemens Digital Industries Software to integrate TIM technology into Siemens’ MindSphere product, an industrial IoT-as-a-service solution. MindSphere customers now have a single dashboard through which they can analyze IoT data, and business users can develop their own data models. This gives them a better understanding of their operations and helps them make smarter decisions as a result. “Thanks to Tangent Works and the ability to use AI and machine learning to automate predictive analytics, even citizen data scientists can easily analyze data and get immediate insights at scale,” says Raymond Kok, senior vice president of cloud application solutions at Siemens Digital Industries Software. “This puts the power of IoT data in the hands of every user.”

The team at Tangent Works is now developing ways to add automated modeling functions to existing AWS machine learning and visualization tools. It is also refining the company’s time series anomaly detection technology so it can be applied to the entire lifecycle of machine learning development and operations. This means the system would be able to monitor itself and automatically improve its own modeling, providing customers with faster access to data models.

In competitive markets, making good decisions based on insights derived from machine learning can be the difference between success and failure. Tangent Works brings these advanced analytics capabilities within the reach of every organization. “Using AWS, we help businesses realize the benefits of machine learning. And they don’t need a dedicated data science team to do it,” says Van Santvliet.