APEX Developers

Quantum — Analytics

Status: Preview · Spec: openapi/quantum.yaml · Base URL: local/companion deployments only (port 9130) — no cloud deployment yet

Quantum is in Preview: analytical outputs are demonstration-grade in the current release, and authentication hardening is in progress. The contract shapes are stable to build against — integrate behind a flag and validate outputs with your platform contact before relying on them.

Quantum runs statistical and machine-learning analysis over structured data: anomaly detection, clustering, correlation, trend decomposition, network analysis, and model-based prediction.

Auth & tenancy: Bearer credential; tenant via the X-Tenant-ID header (missing → 400 MISSING_TENANT, unknown → 404 UNKNOWN_TENANT). Request fields use camelCase; response casing varies by operation (prediction responses camelCase; analysis, pattern, and model-catalog responses snake_case — follow the spec schemas). Responses echo X-Request-ID and X-Response-Time.

The surface

Operation What it does
POST /api/v1/analyze One-shot analysis: {analysisType, dataset, entity?, parameters?, includeVisualization?} where analysisType ∈ anomaly_detection, clustering, correlation, trend_decomposition, network_analysis. Returns findings[], a summary, optional visualization_data, and vector_confidence.
POST /api/v1/patterns Cross-entity pattern detection over entities, dimensions, and a timeframe.
POST /api/v1/predict Model-based prediction: {model, inputs, horizon?, includeExplanation?}; unknown model → MODEL_NOT_FOUND.
GET /api/v1/models (+ /{model_name}) The available model catalog and per-model detail — discover, never hardcode.
GET /health Liveness.

Errors are coded: VALIDATION_ERROR (422), INTERNAL_ERROR (500), and MODEL_NOT_FOUND — the current release delivers MODEL_NOT_FOUND in the error envelope with HTTP status 200, so check the ok field rather than relying on the status code.

Integration notes