Detect anomaly automatically

As soon as you send your metrics, Anomaly will start learning from it. After this learning process done, it will be able to detect unusual patterns when they occurred. It’s a smart detection that adapt itself based on your metrics.


learn from metrics


To start the learning process simply send your metrics to Anomaly. Our algorithms will never stop learning from your data. As a result, it will become more accurate over time. At first it will notice simple patterns such as minimum and maximum. With time, it will start detecting much complex patterns such as trends, correlation, patterns and many more… Simply send your metrics Anomaly will learn from it.
To achieve this learning we use cutting edge techniques such as machine learning, supervised learning, neural networks, pattern recognition as well as our own mathematical model.
To learn the most of your metrics, we try every possibility. We transform each of your metric into million others by smoothing, aggregating… And try to learn from them. We even compute transformed metric again each other’s creating trillion or quadrillion possibilities.

detect anomaly


Using all the patterns learnt from your metrics it’s possible to detect unusual behaviors. We watch in real-time every single metric’s points you send us. If it doesn’t match our filters we trigger an alert. Based on the learning mechanism the detection is more or less flexible to anomalies. It depends on how many times our forecast were validated. When the prediction were correct million of times the detection is very strict to trigger alert. On the opposite, if the prediction was correct only a few hundreds of times the alert will be more flexible to the deviation.
Detection is useless if you get don’t get the information or if you get it too late. That’s why we notice you by email, SMS or other alerting services such as PagerDuty. Alerting is a difficult subject, it needs to be easy to understand in order to take quick action. We work hard to deliver understandable alerting in plain English.

improve detection


If you’re maintaining your system it will be detected as an anomaly. But is it really an anomaly? Technically -yes- but from your point of view it isn’t. Anomaly is your third eyes not your brain. It needs to be told when an anomaly is detected correctly or not. It will include your telling into future detection and learning. On a similar way, it might find patterns who aren’t relevant to you. Maybe just a coincidence, trends who aren’t or correlation that don’t make sense. When you get a false alert simply tell us what to do for the future. The more you tell Anomaly what make sense to you the more accurate it will become.

Detect anomaly and monitor your metrics now.

Sign up for Anomaly today.