Anomify

With a free account, you can send metrics from any source to a dedicated endpoint via POST request. You can also integrate directly using Telegraf, InfluxDB, Graphite, or Prometheus. Depending on your service level, various plugins and connectors are available for tools such as CollectD, StatsD, Google Analytics, OpenTelemetry Collector, MySQL, and SQL Server. If your datastore is not currently supported, please contact us to request its addition to our roadmap.

Anomify adds a new dimension to your observability stack. While traditional tools rely on manual rules and thresholds that often exclude most metrics, Anomify monitors all your metrics in the background to identify abnormal changes. This provides deeper insights into your systems. Users can train the system to recognize expected patterns, which reduces false positives. Unlike most platforms that use opaque unsupervised learning, Anomify offers transparent supervision, providing human-readable explanations for its predictions that can be adjusted to match your system's expected behavior.

1. Sign up for a free account (2 minutes). 2. Send metrics to the ingestion endpoint or connect your time-series database (30 minutes). 3. Observe as your metrics are continuously analyzed for anomalies and correlations. 4. Configure alerts for your most important metrics. 5. Train the analysis with a single click when false positives occur.

Yes, you can manage your metrics directly through the dashboard or via the API.

Every new metric sent to Anomify counts toward your quota. If you stop sending data for a specific metric, it will still count against your quota until you delete the metric to free up space.

Anomify tracks your metric volume and will drop any data sent exceeding your quota. You can use a test header when sending metrics to bypass analysis and avoid committing them to your quota.

Yes, please contact hello@anomify.ai for details.

There is no upper limit on the number of metrics. We can scale horizontally to accommodate your requirements.

Storage depends on the data source. Anomify is not a primary metric store; it preprocesses raw data every minute for analysis and then discards the raw values. Consequently, recorded values may differ slightly from your source store, though trends remain consistent. Data is stored at a frequency of one point per minute, with retention periods ranging from 30 days to 2 years depending on the source. We store trained time-series segments indefinitely for pattern matching. Your data is never shared with third parties.

Anomify requires approximately 7 hours of data to begin analysis and allow for basic test alerts. However, we recommend waiting for 7 days of data to ensure the system has a comprehensive understanding of your expected data ranges before setting up production alerts.

Training the machine learning process helps you tune Anomify to recognize your system's normal behavior, receive alerts on unexpected patterns, reduce false positive notifications, and gain deeper visibility into your metrics.

Billing cycles are based on the calendar month, starting on the day you subscribed to a paid account.

Your account will be updated immediately, and you will be charged a prorated amount for the remainder of the cycle. If you are downgrading, the prorated charge will be $0.00. Your billing cycle date will remain unchanged.

Cancellations take effect at the end of your current billing cycle, which may be between 1 and 30 days from the date of cancellation. After this period, your account will revert to the free plan.