Streem.ai: The AI Magnifying Glass for Engineers

Streem.ai: The AI Magnifying Glass for Engineers

Avi Elran, Chief Executive Officer, Streem.aiAvi Elran, Chief Executive Officer A twice-told tale by business thought leaders is that “the faster and better a company makes decisions, the greater the outcomes”. But given the rapidly changing, data-driven, business environment, how do organisations remain relevant when their ability to generate data outruns their ability to analyse it?

Particularly in the field of manufacturing, it has undoubtedly become a matter of concern for manufacturers that are unable to make these quick decisions. An amalgam of the big data explosion, an increasingly complex and connected ecosystem, and the pressing desire to stay at the front line has evoked the need for organisations to use data analytics for effective decision-making. Manufacturers are under constant pressure to bring new and sophisticated products to market quickly while maintaining product quality. But the rising tide of sensor data is fast becoming a gold mine of missed opportunities, rendering it nearly impossible for engineers to analyse all the data to identify anomalies and keep a close watch on the production systems.

Without the right analytics tools to derive insights, manufacturers, today, find themselves lost in a sea of unlabeled and constantly changing data. In this dynamic environment, applying the current status quo, rule-based systems and even the more sophisticated supervised machine learning tools that require historical and pre-labeled data is not possible. The future of manufacturing technology depends on AI and big data solutions that equip manufacturing firms with the inspection capabilities to make sense of this sea of unlabeled data, and rid engineers of the expensive and time-consuming task of manually deriving labels.

Bridging the gap between the unused and unlabeled data and the insights that manufacturers seek from this data is Streem.ai. A revolutionary AI firm, Streem.ai offers a solution that combines unsupervised anomaly detection with robust anomaly classification, which simplifies previously unknown errors to enable better decision-making in production systems. “Streem.ai doesn’t require extensive labeling. It detects abnormal patterns and then classifies those,” says Avi Elran, CEO of Streem.ai.

The solution is specifically designed so that manufacturers can handle the data using an unsupervised machine learning algorithm that helps detect patterns and abnormalities in all the new products’ data in a very short period of time. “We enable engineers to unlock the hidden potential of their product data using machine learning algorithms,” says Gary Abela, CCO of Streem.ai.

Not requiring historical data or labels is a key distinguishing factor that allows Streem.ai to onboard any machine generated sensory data set. After establishing the underlying patterns and discovering abnormal behaviours, Streem.ai then applies further algorithms to extract insights from this new knowledge that it discovered in the data.


Engineers are the experts, we are the enablers


In essence, augmenting the information it discovers, and creating additional value for its users.

In a League of its Own

After two years of rigorously testing and developing solutions across the full spectrum of available algorithms, the company deduced that the utilisation of unsupervised machine learning in industrial data drives the most value out of their technology. “Streem.ai’s unique approach enables engineers and subject matter experts to perceive data from their systems in a new light,” says Abela. By allowing them to tag, label, and interact with the patterns, the underlying algorithms continuously improve and enhance the quality and relevance of the result.

There are many sectors within manufacturing that Streem.ai can work with. As Elran states, “We are not domain experts. We are data experts.” Their artificial intelligence system can consume any repetitive sensor data. Examples include pipeline leak detection, hardware/software-in-the-loop simulations, and shop floor processes as well as assembly line data, test bench data. Streem.ai takes this direction because they discovered that patterns in the data could be more reliably understood from repetitive sequences, which allows them to develop a more stable model of what is normal.

Abela notes that while data scientists and engineers possess the expertise to work on the data sets, there is a gap in effectively integrating that domain expertise into decision support systems. Streem.ai steps in to bring that domain knowledge to the forefront of the AI system, enabling engineers to give the context in a more automated fashion.

Unlike most existing or new solutions used by manufacturers that require data science teams in organisations, Streem.ai’s solution stands unique. “Our unsupervised approach ensures that no new behaviour or anomalies are missed, thereby ensuring a robust and reliable solution,” says Abela.

It’s what you do with the Data

The main benefit that companies have with Streem.ai is the significant reduction of time that engineers spend on data inspection, which they can instead use for engineering tasks geared toward improving quality and focusing on developing better products, rather than on eyeballing data and inspection.

Webasto, a global tier 1 supplier of heaters and sunroofs to the automotive sector and a true leader in innovation, experienced the benefits of Streem.ai firsthand when they found themselves swamped with the quantity of data and its complexity. Before delivery, the client’s heaters undergo extensive inspections and test runs—a laborious and lengthy process of producing daily sensor recordings from various devices.

Abou George, Chief Operating Officer and Gary Abela, Chief Commercial Officer
With their time to market at stake, Webasto’s engineers were required to monitor this data as fast and in as little time as possible. They found that Streem. ai’s tool digests all the complex data to detect patterns to identify abnormal behaviour, which helped reduce the time engineers spent on data analysis and monitoring by 85 percent.

"Streem.ai doesn’t require extensive labeling. It detects abnormal patterns and then classifies those"

There’s more to what Streem.ai can do than meets the eye. The company has been witness to the increasing pressure to get products to market faster, and the consequent growth in simulation, as a validation technique, only skyrockets the data being produced further. To prevent manufacturers from drowning in this data, Streem.ai understands the patterns and deviations from manufacturers’ expectations in near real-time and notifies engineers of potential bugs in the system.

The Engineer’s Companion

The beauty of what Streem.ai does is that the solution can fit across different divisions within a manufacturing company. The idea is to operate within an organisation and be the vendor of choice across different data sets in different divisions. That’s the company’s “land and expand” approach with every customer that they come across.

Streem.ai is currently focused on German and nearby European markets with an added interest to continue expanding internationally in the future. The plan, for now, is centred on enhancing the ability to automatically find and extract patterns, and help companies navigate through their data by applying smarter and more advanced algorithms.

Despite Streem.ai being an early version, it has received tremendous traction and interest from customers that are focused on AI solutions. Unlike most AI solutions, Streem.ai can provide quick results based on data already collected by manufacturers without having to go through complex integration processes. This means customers can validate the solution with minimum cost and effort from their side before moving to full-scale implementation.

Elran asserts that the goal of Streem.ai is not to replace engineers. Their vision is to empower engineers with a valuable tool to augment their ability and knowledge. “Engineers are the experts, we are the enablers,” says Elran. The company wants to help engineers and manufacturing companies around the world gain a holistic understanding of the current state of their machines. Elran sums it up with, “We see our tool as the engineers’ AI-driven magnifying glass. As the consumer world now has everyday smart companions in the form of Siri and Google Assistant, Streem.ai will be the engineer’s companion.”

- Ava Gracia
    December 18, 2019
Share this Article:
Streem.ai

Company
Streem.ai

Headquarters
Berlin, Germany

Management
Avi Elran, Chief Executive Officer and Abou George, Chief Operating Officer and Gary Abela, Chief Commercial Officer

Description
Streem.ai offers a home-grown AI solution specifically designed to help manufacturers handle the data using an unsupervised machine learning algorithm and get products out to market as quickly as possible. The solution is specifically designed so that manufacturers can handle the data using an unsupervised machine learning algorithm that helps detect patterns and abnormalities in all the new products’ data in a very short period of time. Streem.ai enables engineers to unlock the hidden potential of their product data using machine learning algorithms. Streem.ai’s unsupervised approach ensures that no new behaviour is missed, thereby ensuring a robust and reliable solution