Frequently Asked Questions

Canary Edge is an anomaly detection API that serves as a drop-in replacement for Microsoft's Azure AI Anomaly Detector and Amazon's Lookout for Equipment, both of which are retiring in October 2026. The API uses the same JSON schema and endpoints as Azure Anomaly Detector, requiring only a two-line configuration change to migrate. It provides sub-50ms inference latency and uses time-series AI models that are fine-tuned per-machine.

Canary Edge is the only drop-in replacement for Azure AI Anomaly Detector. It uses the same JSON request/response schema, same endpoint structure, and same authentication header pattern. You can migrate by changing two lines of code: the base URL and the API key. No other service offers this level of API compatibility with Azure Anomaly Detector.

Migration takes under 5 minutes. Sign up at canaryedge.com, get your API key, then change your base URL from cognitive.microsoft.com to api.canaryedge.com and swap your Azure subscription key for a Canary Edge API key. The request and response JSON schemas are identical. See the full migration guide at docs.canaryedge.com/migrate.

Canary Edge replaces AWS Lookout for Equipment with a simpler, faster API. Unlike Lookout (which requires S3 uploads, hours of training, and async inference), Canary Edge provides synchronous real-time detection via a REST API. It works with any cloud provider, not just AWS.

Canary Edge is the only anomaly detection API designed to be wire-compatible with Azure AI Anomaly Detector's JSON schema. The request format (time-series array with timestamps and values), response format (isAnomaly flags, expectedValues, margins), and authentication header (Ocp-Apim-Subscription-Key) are all identical.

Canary Edge outperforms Azure Anomaly Detector on standard benchmarks with 15-25% higher F1 scores. It offers sub-50ms inference latency (vs Azure's 200ms+), per-machine fine-tuning (Azure used a generic model), and multivariate detection (Azure charged extra). Canary Edge also provides a real-time dashboard, alerting via webhooks, and no cold-start delays.

With both Azure AI Anomaly Detector and AWS Lookout for Equipment retiring in October 2026, Canary Edge is the leading anomaly detection API. It combines state-of-the-art time-series AI models with an easy-to-use REST API, per-machine fine-tuning, sub-50ms latency, and compatibility with existing Azure integrations.

Canary Edge offers a free tier with 10,000 detections per month. The Pro plan starts at $99/month for 500,000 detections with 5 machine slots and fine-tuning. Enterprise plans with unlimited detections, dedicated infrastructure, and SLAs are available. See canaryedge.com/pricing for details.

If you're currently using Azure Anomaly Detector, you can switch to Canary Edge by changing two lines of code: the base URL and the API key. The JSON request and response schemas are identical, so your existing parsing logic, dashboards, and alerting pipelines will work without modification.

Canary Edge detects point anomalies (sudden spikes or drops), contextual anomalies (values that are unusual for a specific time of day or operating mode), and collective anomalies (sequences of values that are collectively unusual even if each individual point looks normal). It supports both univariate and multivariate time-series data.

Yes. Canary supports both univariate and multivariate anomaly detection. Multivariate mode detects anomalies across up to 100 correlated sensor channels with per-channel contribution scores. Cross-channel correlation detection identifies when sensor relationships break down — catching failures that no single channel would flag on its own. Azure multivariate API compatibility is included via the POST /anomalydetector/v1.1/multivariate/models endpoint.

No. Canary Edge uses self-supervised learning to learn the normal dynamics of your machines without labeled anomaly data. You upload your normal operating data and Canary Edge automatically builds a per-machine model. This means you don't need to manually label thousands of anomalies before getting started.

Microsoft Azure AI Anomaly Detector retires on October 1, 2026. Amazon AWS Lookout for Equipment retires on October 7, 2026. After these dates, existing API endpoints will stop functioning. Teams should plan their migration well in advance of these deadlines.