An AI assistant using proprietary unsupervised algorithm(s) to automatically analyse large volumes of complex data to detect anomalies and recognize patterns in real time

Project Information

Streem.ai

Grant agreement ID: 960719

DOI

10.3030/960719

Project closed

EC signature date 25 June 2020

Start date 1 July 2020

End date 30 June 2022

Funded under

• INDUSTRIAL LEADERSHIP - Innovation In SMEs

Total cost

€ 2 591 250,00

EU contribution

€ 1 813 875,00

Coordinated by

STREEM AI GMBH
Germany

Periodic Reporting for period 2 - Streem.ai (An AI assistant using proprietary unsupervised algorithm(s) to automatically analyse large volumes of complex data to detect anomalies and recognize patterns in real time)

Reporting period: 2021-04-01 to 2022-06-30

Summary of the context and overall objectives of the project

Streem.ai is ready to help anyone facing the daunting task of monitoring large amounts of time-series data. It acts as a magnifying glass for domain experts, reducing manual data inspection efforts to allow more time for root cause analysis. Streem’s technology helps manufacturers get complex products to market faster and with fewer defects, augmenting domain knowledge with artificial intelligence to locate the earliest indicators of future product failures.

Our vision is to help professionals such as engineers, domain experts, and analysts working with large amounts of data to make better decisions, quickly and easily, and empower humans through smart algorithms. Our intelligent anomaly detection tool maximizes data insights for engineers, analysts, and managers, enabling them to make fast and informed decisions. The overall objectives of the project are to further develop our product and technology, with a focus on scalability and reliability, and begin early commercialization within the manufacturing sector.

Streem.ai’s technology differentiation is its high performance and accuracy along with its ability to learn and continuously improve over time automatically. Our proprietary self-learning algorithm and ability to operationalize ML at scale for almost 70% of the sensory data in manufacturing enable us to quickly onboard customers in less than two weeks. The solution distinguished itself from existing solutions through the following:
Works with unlabelled data (majority of the data in manufacturing is unlabelled)
Designed specifically for domain experts without the need of having a data scientist in the loop. This is a huge benefit to SMEs who do not have in-house data scientists
Customers get their first insights within a week’s time with the system gradually improving over time. Typically, AI systems take months to be deployed
Seamlessly scales across use cases within an organisation making it easy for customers to rely on a unified solution

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

The key results achieved during this reporting period are as follows:
- Completed build of 1st version of an easy and intuitive application for the end-user supported by a scalable and robust backend.
- In process of developing and designing all tools in the machine learning pipeline tightly integrated with infrastructure services for scheduling, monitoring, logging, data management, storage, and security.
- Continuous improvement of Streem.ai core AI engine including research of state-of-the-art algorithms
- Performance enhancements by running pilot tests, incorporating client feedback, and assessing test results.
- Onboarding of early trial customers

The key results achieved after the completion of the project are as follows:
- Easy and initiative dashboard
- Scalable backend
- Self-service data ingestion and processing capability
- Microservice-based backend developed for data storage, processing and inference
- ML pipeline integrated into backend services.
- Continuous Monitoring and logging layer built
- All security compliances have been completed by ensuring the use of compliant cloud services with all data handling-related services
- 7 algorithms have been utilised, optimised and deployed in the ML pipeline
- A novel benchmarking algorithm is deployed in the backend that identifies the best algorithm for a given data set.
- Data pipeline integrity tested and verified
- Performance of algorithms measure and systems in place to identify the best algorithm
- Test customer feedback collected to identify UI improvements and incorporated them into Streem application
- Privacy: In order to comply with customers’ privacy requirements, public and private cloud versions of Streem are available
- Security: Only security-compliant services have been used within the application
- Private customer, user data is managed as per GDPR guidelines

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

Streem.ai is a scalable self-learning AI in the fast-growing $1.25 billion cognitive manufacturing markets in Europe. The product provides manufacturers with up to 67% faster time to market and increases the productivity of engineers by 80%. It helps professionals such as engineers, domain experts, and analysts work with large amounts of data to make smarter and faster decisions. Our intelligent anomaly detection and pattern recognition tool maximizes data insights for these professionals, enabling them to make fast and informed decisions.

In the post-COVID times, as manufacturers are under pressure to produce high-quality products, Streem.ai can aid in this by enabling engineers with an AI-enabled assistant.

Streem.ai is a scalable self-learning AI in the fast-growing $1.25 Bn cognitive manufacturing market in Europe. The product provides manufacturers the following benefits:
- Up to 67% faster time-to-market
- Increase in productivity of engineers by 80%

Socio-econimoc impact:
- Market pressure: In the post-pandemic world, there is a lot of pressure on manufacturers to adapt to the changing demands of consumers, while ensuring high-quality products are delivered to market quickly.
- Resource gap: Large tech giants such as IBM and Amazon are beginning to enable enterprises to develop algorithm-based solutions, however, they require highly skilled data scientists to use them. The bedrock of the modern economy is SMEs, such as Webasto, who do not have the necessary skilled data scientists to become AI-first enterprises. A European Data Market study estimates that the EU will have a shortage of 485,000 data workers in 2020.
Streem.ai provides the power of AI without the need for data scientists.

Streem.ai Logo