Let AI read
your telemetry.
Anomaly detection, Prophet forecasting, natural-language querying, and incident root-cause summaries — applied to your own logs, metrics, and traces. RAG over your runbooks turns past incidents into answers, so you spend less time hunting and more time fixing.
Find the anomaly you didn’t write a rule for
Z-score and IQR detection score every point in your metric time-series and flag what breaks the pattern — no thresholds to maintain, no baselines to hand-tune.
- Z-score and IQR anomaly scoring on any metric series
- Per-point flagging with a configurable sensitivity threshold
- Anomaly-driven alerts wired into the alerting engine
- Works on existing telemetry — no new instrumentation
See the spike before it happens
Prophet-based forecasting projects metrics 24 hours to 14 days out with confidence bands. Plan capacity, anticipate saturation, and budget infrastructure before the page fires.
- Prophet forecasts at 24h, 3d, 7d, and 14d horizons
- Upper and lower confidence bands on every prediction
- Capacity planning for CPU, memory, disk, and request volume
- Forecast any metric or a custom series you provide
Ask in English. Resolve with context.
Translate plain-English questions into safe, parameterized ClickHouse SQL. When an incident hits, AI summarizes the likely root cause from logs and traces and surfaces relevant runbooks via RAG.
- Natural-language to ClickHouse SQL, injection-safe by design
- Schema-aware queries across logs, spans, metrics, and events
- LLM incident summaries with root cause and remediation steps
- RAG over runbooks and past incidents for instant context
Built for every team that cares about reliability
One platform, tailored to how your team actually works.
Faster root cause
AI summarizes the incident and points to the runbook while you’re still reading the page.
Plan capacity
Forecasts and anomaly detection flag saturation days before it bites.
Query without SQL
Ask questions in plain English and get answers from the telemetry directly.
Ask in plain English
Questions you can type instead of queries
Natural-language querying runs against your own telemetry. These are the examples the product ships with.
- How many error logs appeared in the last hour?
- What are the top 5 services by request count today?
- Which traces had duration over 2 seconds in the past hour?
- What's the P95 latency for the checkout endpoint?
- How many 500 errors happened yesterday?
AI features grounded in your data, not a demo
Anomaly detection runs Z-score and IQR statistics over your real metric series. Forecasts come from Prophet models fitted to your history. Natural-language queries compile to parameterized ClickHouse SQL, and RAG retrieves from runbooks you upload — every answer traces back to your own telemetry.
Also in the platform
Start using AI insights in 5 minutes
Free forever for up to 3 services. No credit card required.