Our powerful technology is a result of hard work, customer focus, team spirit and determination Alexander Helmer
Co-Founder & COO
Co-Founder & CTO
We have a
vision to enable the world to conquer operational complexity and make it manageable for people. We are on a mission to develop the world’s most effective tools to resolve machine underperformance Together we can create value from day one Request your Demo today:
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Frequently asked questions
Jungle is committed to helping its clients reach their goals, and maximise their asset performance. Our strong sense of identification with client projects means that we are constantly striving to provide solutions, even for issues they aren’t yet aware of. To this end, we adopt an innovative approach to artificial intelligence and machine learning technology. Our tech has delivered value at GW scale for global leaders in the renewable energy sector.
The asset owner is the owner of the data. Even if the data is on our cloud, we treat it with the utmost privacy, it will remain under your ownership and we will manage it at your request.
What if our OEM doesn't make our data available?
The asset owner is the owner of the data. Throughout our journey with our clients, we have always managed to gain access to the raw data. We have the right experience and knowledge in setting up these types of connections. It is essential to have the right credentials in order not to face any problems.
Do you use artificial intelligence?
Yes, our machine learning models are powered by cutting-edge technology. These are specially designed from their core to extract insights from massive datasets composed of thousands of sensors from multiple assets. Our team consists only of experts in machine learning solutions for multivariate time-series generated by electro-mechanical assets.
How much historical data do you need?
It is preferable to get at least 12 months worth of data
to cover a full seasonal cycle. The more historical data we feed into the machine learning models, the better they get. However, we can still get started with a few weeks of data.
We usually use ODBC SCADA 10-minute or higher resolution (up to 1 second), including sensor and event data.
Do you need annotated historical failures?
No. We use unsupervised machine learning on unlabelled data sets. Our normality models only need to be trained with the available historical SCADA data with no annotated historical failures.
Which components do you monitor?
Our normality models can learn the dynamics of all electro-mechanical components that have at least one sensor. Canopy then tracks actual versus expected behaviour in real time for these components.