Observability & AI-Ops
aiAxonIQ
OpenTelemetry Observability on ClickHouse, Built for High Cardinality
Per-host and per-custom-metric pricing makes high-cardinality data such as user IDs, tenant IDs, and request paths expensive to keep, so teams drop the labels they need most during an incident. aiAxonIQ stores telemetry in ClickHouse, where cardinality costs disk and compute rather than a line item, and compresses it column by column. Go collectors stream OTLP data through a Kafka buffer that absorbs ingestion spikes, Prophet models forecast latency and capacity trends, and NL→SQL turns a question into ClickHouse SQL you can read and edit.
Key metrics
Design targets and benchmarks, validated with each customer during a pilot.
Capabilities
OpenTelemetry-Native Ingestion
Go workers receive OTLP logs, metrics, and traces and write them to ClickHouse, with OpenSearch available for full-text log search.
Natural Language to SQL (NL→SQL)
Ask “show error spikes in checkout during the last deploy” and get the generated ClickHouse SQL alongside the result.
Forecasting & Anomaly Detection
Seasonality-aware Prophet forecasts flag latency regressions and capacity trends before they breach an SLO.
Multi-Tenant Isolation
Role-based access control and tenant-scoped ClickHouse partitions for shared and dedicated deployments.
Distributed Trace Waterfalls
Span-level views that show which microservice hop added the latency, across Kubernetes and cloud services.
High-Cardinality Columnar Storage
Keep user, tenant, and request-level attributes on every span without per-series billing.
Architecture
- Go/Gin API with ClickHouse columnar storage
- Kafka streaming buffer between collectors and storage
- Python ML service for Prophet forecasting and incident summaries
- Next.js dashboards, plus Grafana and Prometheus integrations
Business Outcomes
- Targets 60%+ lower observability spend than per-host SaaS monitoring pricing
- Flags regressions from forecast trends, not only static thresholds
- Lets any engineer query telemetry in plain English or SQL
Who It's For
VP of Site Reliability / Head of Infra
Replaces per-host APM pricing with storage you size and control.
DevOps / Incident Commander
Gets from alert to likely cause faster with correlated traces, forecasts, and NL→SQL.
Staff Backend Engineer
Queries raw spans and logs in SQL instead of pre-aggregated dashboards.
Integrations
- OpenTelemetry (OTLP)
- ClickHouse
- Apache Kafka
- OpenSearch
- Prometheus
- Grafana
- Kubernetes
- AWS CloudWatch
Works With
Application Security & Non-Human Identity (NHI)
TrustIvan
Container and SBOM scanning, machine-identity inventory, and runtime authorization for AI agents
Enterprise AI & Agent Orchestration
MITHUNAI
Private AI assistants and multi-agent workflows that cite their sources
Open-Source Continuous Compliance (GRC)
trustlynce
Open-source compliance automation with 270+ MCP tools for evidence, controls, and audits