> ## Documentation Index
> Fetch the complete documentation index at: https://docs.canaryedge.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Migrate from Azure Anomaly Detector

> Switch from Azure AD to Canary Edge with minimal code changes

Azure Anomaly Detector retires **October 1, 2026**. Canary Edge is API-compatible — in most cases, you only need to change the endpoint URL and authentication header.

## Step 1: Get Your Canary Edge API Key

Sign up at [canaryedge.com](https://canaryedge.com) and generate an API key from the dashboard.

## Step 2: Update Your Endpoint

Replace the Azure endpoint with Canary Edge:

```diff theme={null}
- https://{your-resource}.cognitiveservices.azure.com/anomalydetector/v1.0/timeseries/entire/detect
+ https://api.canaryedge.com/v1/timeseries/entire/detect
```

## Step 3: Update Authentication

Canary Edge accepts the same `Ocp-Apim-Subscription-Key` header that Azure uses:

```diff theme={null}
  headers = {
      "Content-Type": "application/json",
-     "Ocp-Apim-Subscription-Key": "YOUR_AZURE_KEY",
+     "Ocp-Apim-Subscription-Key": "YOUR_CANARY_EDGE_KEY",
  }
```

## Step 4: Verify

Send the same request you were sending to Azure. The response format is identical:

```json theme={null}
{
  "isAnomaly": [false, false, true, false],
  "isPositiveAnomaly": [false, false, true, false],
  "isNegativeAnomaly": [false, false, false, false],
  "expectedValues": [10.5, 10.8, 10.6, 10.7],
  "upperMargins": [2.1, 2.16, 2.12, 2.14],
  "lowerMargins": [2.1, 2.16, 2.12, 2.14],
  "period": 0
}
```

## Endpoint Mapping

| Azure Endpoint                   | Canary Edge Endpoint                |
| -------------------------------- | ----------------------------------- |
| `/timeseries/entire/detect`      | `/v1/timeseries/entire/detect`      |
| `/timeseries/last/detect`        | `/v1/timeseries/last/detect`        |
| `/timeseries/changepoint/detect` | `/v1/timeseries/changepoint/detect` |

## What's Different

* **Better accuracy** — Canary Edge's advanced engine catches anomalies that threshold-based methods miss
* **No resource provisioning** — No Azure resource groups, no Cognitive Services setup
* **Per-machine fine-tuning** — Optional model customization via `/v1/train`
* **Transparent pricing** — Pay per API call via Stripe, no Azure commitment required

## Need Help?

Contact [support@canaryedge.com](mailto:support@canaryedge.com) for migration assistance.