Aether Telemetry

The problem

OpenTelemetry does technical observability — which call took how long. But it doesn’t answer what that behavior means for the system’s contract. In regulated sectors (banking, healthcare, aerospace) the difference between “API responded in 200ms” and “semantic contract violated in 200ms” is everything.

What Aether does

Aether Telemetry observes the real behavior of critical systems against declarative semantic contracts evaluated at the edge (WASM at C2). When a contract is violated, the event reaches the pipeline with full contractual context — not just the metric.

• eBPF for non-intrusive production capture.
• WASM to evaluate rules at the edge without restarting services.
• SPIFFE/SPIRE for cryptographic identity of every observed workload.
• Redpanda as the event bus.
• ClickHouse for sub-second analytical query.

Current state

Feature Freeze v1.0 reached. 546 tests passing. Waiting on deployment directives for the first pilot. Three betas in the pipeline.

When it applies

• Teams running OpenTelemetry but with no way to detect contractual violations.
• Regulated sectors where “everything green in Datadog” is not enough evidence for audit.
• Products where the cost of a semantic deviation is measured in regulation or lives, not clicks.

When it does NOT apply

• Marketing or simple e-commerce sites. No semantic contracts to watch.
• Teams without OpenTelemetry yet — start there first.

Stack

Rust · tokio · eBPF · WASM (wasmtime) · Redpanda · ClickHouse · SPIFFE/SPIRE.