Product

SupPulse

Unified observability, correlation and scheduled inspection: metrics, logs, traces, alerts and change context in one console, so every investigation has evidence behind it.

Book a demo Contact us

A metrics dashboard on screen. — Photo Stephen Dawson · Unsplash

Metrics, logs, traces and alerts live in separate systems, so investigations mean switching tools and losing context. Duplicate alerts bury the events that matter and wear out the on-call rota. Root-cause analysis depends on individual experts, keeping MTTR high. Routine inspection is done by hand, with inconsistent scope and conclusions that are never kept.

SupPulse connects metrics, logs, traces, alerts and change context and lets you query, drill down and correlate them from one console. AI-assisted diagnosis calls the observability tools on demand, collects evidence step by step and forms a root-cause hypothesis; scheduled inspection checks the connected infrastructure on a plan, produces a report you can revisit and pushes it to the team's messaging channel. Existing data sources and collectors stay in place; no observability store has to be migrated.

Capabilities

AI-assisted diagnosis and Q&A: start an investigation in natural language; an agent calls the observability tools, gathers evidence and forms a root-cause hypothesis. Multiple models, expert skills and MCP integrations are configurable; session history and an agent audit trail are kept
Unified observability: query metrics (PromQL), logs (Elasticsearch / Loki), traces and dashboards from one console; APM shows service topology and RED metrics, with drill-down to traces and spans
Scheduled infrastructure inspection: plans by cron and time zone, custom scope, run-now; reports are retained and pushed to Feishu, DingTalk, WeCom or a webhook
Anomaly detection and alert correlation: dynamic baselines and alert rules flag anomalies; rule labels link upstream and downstream alerts, and a wait window suppresses downstream notifications; observe-only, enforce and emergency-release modes, with every decision auditable
Event centre and operations graph: ingest change and runtime events from GitLab, Jenkins, ArgoCD, Kubernetes and more, with search and a timeline; with the optional graph (Neo4j) enabled, the AI can query service dependencies and recent changes through read-only tools
Edge collection and platform governance: OpenTelemetry Collector and OpAMP manage collection nodes; plugins cover hosts, middleware, network and databases; multi-tenancy, IAM, audit and configuration versioning are built in

Connect the tools you have instead of replacing them

One console ties together data queries, event context and AI workflows, respecting the observability investment already made and leaving room to extend. The observability stores and the event centre remain the source of fact; the operations graph adds relationship context; the AI works only through authorised tools, with tenant boundaries, tool permissions and an agent audit trail covering the whole process.

Unified workbench: observability queries · APM · alert management · event centre · AI Q&A · scheduled inspection
Platform services and AI workflows: unified query · label correlation · inspection scheduling · tool-based evidence · permissions and audit
Observability data and operations context: Prometheus / VictoriaMetrics · Elasticsearch / Loki · Jaeger · PostgreSQL event records · Neo4j operations graph (optional)
Collection and event ingestion: OTel Collector · OpAMP · collection plugins · GitLab / Jenkins / ArgoCD / Kubernetes / standard webhooks

A workable way in for each role

From on-call response to daily inspection, from application performance to platform governance, SupPulse covers the operations team's main scenarios.

SRE · reliability: when a host goes offline and MySQL becomes unavailable, host labels link the upstream and downstream alerts to cut duplicate notifications, then the AI queries the observability evidence to help judge the cause
Developers · performance: view service topology and dependencies in APM, drill down to traces and spans, and analyse count / rate / error / P99 per endpoint to locate slow requests and error sources
Platform · administrators: divide responsibilities by tenant, user group and permission; platform and agent audit trails record key actions; configuration versions can be traced back
Operations leads · daily inspection: set a weekday inspection plan over the connected hosts, databases and network devices; reports keep the conclusions and evidence and are pushed to Feishu, DingTalk or WeCom

Observe → correlate → analyse → recommend

Alerts trigger investigations; schedules drive inspections. SupPulse turns scattered signals into evidence that can be checked, and leaves the final remediation decision with the team. AI conclusions assist judgement; their quality depends on the connected data, the model and the workflow configuration. The platform does not promise automatic remediation.

Observe: ingest metrics, logs, traces and change events; keep the source of fact
Correlate: label rules reduce duplicate notifications; dependencies and changes add context
Analyse: an agent calls tools on demand and produces a root-cause hypothesis or an inspection report
Recommend: evidence and suggestions go to the team; people confirm and act

Interface

From signal to evidenceCONTEXT → INSIGHT

metricslogstracesalertschanges

01Observe in one place, keep the scene
Service behaviour and anomaly signals in one console
02Correlate context, narrow the search
Label links, service dependencies and recent changes
03AI gathers evidence, suggests next steps
Root-cause hypothesis, cited evidence, verification steps

Beyond incidents: inspections on a schedule, reports delivered

Signals

3in 1

metrics · logs · traces, one query

Correlation

Linksuppress

label match · wait window · audit

Root cause

AgenticRCA

tool evidence · hypothesis · advice

Inspection

Cronreports

custom scope · history · push

FAQ

Does it intercept existing alert notifications?: Correlation and suppression apply only to alert notifications that pass through the platform's unified entry point; Nightingale's native notification path is not intercepted.
Is the operations graph required?: No. The graph is an optional capability that needs Neo4j; the event centre keeps the factual record and does not decide root cause on its own.
Will the AI fix incidents automatically?: No. The AI produces evidence, a root-cause hypothesis and suggested verification steps; remediation is confirmed and carried out by people. Results depend on the connected data, the model and the workflow configuration.
Do we have to migrate our observability stores?: No. Adapters unify the query results; Prometheus, Elasticsearch, Loki, Jaeger and the rest stay where they are.