Anomify

Anomify is a real-time anomaly detection platform for metrics designed to surface changes across your metric population and speed up root cause analysis

Overview

• Anomify continuously analyzes your metrics, ingesting data and learning normal patterns to detect unusual behavior in real time. It provides insights to help reduce false positives, identify optimization opportunities, and resolve issues faster.

How it works

• Real-Time Analysis: After data ingestion, Anomify learns normal patterns, detects anomalies as they occur, and continuously improves the model.
• Cutting Edge Detection: Combines supervised ML, pattern matching, correlation, and custom algorithms to cover all anomaly types.
• Training and Personalization: Users can train Anomify to recognize expected patterns, reducing false positives and aligning predictions with your mental model.

What it helps with

• Reducing false positive alerts by using a consensus of analyses and recognizing seasonal patterns.
• Resolving issues quickly by listing metrics that changed unexpectedly at the time of an event.
• Finding optimisation opportunities by surfacing abnormal metrics that may reveal optimization ideas.

Who it’s for

• Managers: Provides peace of mind about system performance.
• Site Reliability Engineers: Helps manage downtime and optimize error budgets with deep event insights.
• DevOps engineers: Supports evolving DevOps practices and reduces pager fatigue.

Supported integrations and data sources

• Connects to common time-series databases and tools via plugins/connectors (details below).
• Can receive metrics from various sources using ingestion endpoints or through TSDBs like Prometheus, InfluxDB, Graphite, and others.
• Plugins/connectors for: WordPress, Microsoft Teams, Email, Webhook, Slack, Open Telemetry, Datadog, and more (with additional integrations planned).
• HTTP API, Telegraf, and other standard metric collection methods are supported; Datadog/Graphite/Prometheus/Open Telemetry connections are highlighted in the integrations list.

How to get started

• Sign up for a free account.
• Send metrics to the ingestion endpoint or connect a TSDB.
• Monitor anomalies and correlations, then set up alerts on the most important metrics.
• Train the system to reduce false positives and improve alignment with your expected behavior.

Billing and data handling

• Billing cycles are monthly. Plans can be upgraded/downgraded with prorations as needed.
• Data storage depends on the data source; Anomify preprocesses raw data every minute and stores processed data for analysis, with raw values discarded. Data retention varies by source (e.g., 30 days to 2 years).
• Data is not shared with third parties.
• After ~7 hours of data, Anomify can start providing useful analysis and alerts; for robust alerts, ~7 days of data is recommended.

Training and optimization

• Training the system tunes Anomify to your environment, reducing false positives and delivering more relevant alerts.
• You can add or remove metrics at any time; metrics count against your quota unless you delete them.

• Real-time anomaly detection for metrics across a population
• Learn normal patterns and detect unusual behavior in real time
• Consensus-based approach to reduce false positives
• Seasonal pattern recognition and ML-driven improvement
• Quick root-cause analysis with pre-compiled change-detection metrics
• Alerts and correlations to help with incident response
• Flexible ingestion options: endpoints, Telegraf, Prometheus, InfluxDB, Graphite, etc.
• Wide range of integrations including Slack, Microsoft Teams, Email, Webhook, OpenTelemetry, Datadog, WordPress, and more
• Scalable horizontal architecture with no upper limit on metrics

Notes

• If you’re evaluating, it’s recommended to start with a free account and gradually onboard key metrics to observe anomaly detection and alerting behavior before expanding usage.