Free Anomaly Detection
Online — No Signup

Paste your data, click scan, see anomalies instantly. No machine learning, no model management, deterministic results every time. A drop-in Azure Anomaly Detector replacement.

Azure AD ReplacementAny Data TypeOne API CallDeterministic

Up to 50 samples · Comma, space, or newline separated

Up to 100 samples · Or upload a CSV below

Anomaly Detection API — Ready for Your Pipeline

One API call. No model management. Deterministic results every time. Works with JSON, text, images, time series — anything.

Azure Anomaly Detector retires October 2026 — switch to WaveGuard now.

Get API Access
Free tier included
Python SDK
Pay-per-call pricing
Interactive docs
Deterministic output
Any data type

How Physics-Based Anomaly Detection Works

01

Encode

Your data is mapped onto a 3D computational grid. Each sample becomes initial energy in a wave field.

02

Evolve

A physics simulation runs 150 time-steps. Normal data produces stable wave patterns; anomalies create chaotic fields.

03

Detect

Wave-field fingerprints are compared statistically. Anomalies stand out because the physics responds differently to unusual inputs.

WaveGuard vs Azure Anomaly Detector — Feature Comparison

FeatureAzure ADWaveGuard
Data formatsTime series onlyAny (JSON, text, numeric, images)
Minimum data12 data points2 samples
API calls needed2+ (create model → detect)1 (single scan)
State managementYou manage models & lifecycleStateless — nothing to manage
DeterministicNoYes — same input = same output
Pricing$0.314 / 1K transactionsFree tier + pay per call
StatusRetiring Oct 2026Active & growing

One API Call — That's It

curl -X POST https://waveguard.p.rapidapi.com/v1/scan \
  -H "Content-Type: application/json" \
  -H "X-RapidAPI-Key: YOUR_KEY" \
  -d '{
    "training": [1.0, 1.1, 1.2, 1.0, 0.9],
    "test":     [1.1, 5.7, 1.0]
  }'

Returns per-sample anomaly scores, confidence, and feature-level explanations.

Frequently Asked Questions

What is WaveGuard?
WaveGuard is a physics-based anomaly detection engine. It maps your data onto a 3D computational grid, runs a wave-equation simulation, and detects outliers from the resulting field fingerprints — no machine learning, no model training required.
How does WaveGuard compare to Azure Anomaly Detector?
WaveGuard is stateless (no model lifecycle), works with any data type (not just time series), needs only 2 baseline samples (vs 12), and gives deterministic output. Azure Anomaly Detector retires in October 2026; WaveGuard is a drop-in replacement.
Is the free anomaly detection tool really free?
Yes. The web tool allows unlimited scans of up to 50 baseline and 100 test samples per request at no cost, with no signup. For higher limits and programmatic access, paid plans are available on RapidAPI.
What data formats does WaveGuard support?
WaveGuard accepts any numeric data — time series, sensor readings, financial data, IoT telemetry, server metrics, or any CSV/JSON array of numbers. Future versions will support text and image inputs.
Do I need to train a model?
No. WaveGuard is completely stateless. Send your baseline (normal) data and test data in a single API call. There is no model to create, train, update, or manage.
How fast is the anomaly detection?
Each scan typically completes in 200–600 ms on the free tier. The physics simulation runs on GPU, so even large datasets are processed quickly.