Ask your network anything
"Which links are saturating?" "What changed right before this outage?" Get answers in plain language, grounded in live topology, telemetry, and logs.
AI Ops
Ask your network. Get answers you can trust.
ServiceRadar is built to be operated by AI. Ask questions in plain language, let it surface anomalies and forecast capacity on its own, and connect the assistants your team already uses through an open Model Context Protocol server.
Demo login:
demo@localhost
·
serviceradar
This isn't a bolt-on chatbot. ServiceRadar's AI reasons over your live network: its real topology, telemetry, and OCSF-normalized security signal. It shows you the exact query behind every answer, so you get results you can trust, act on, and automate.
From a question to an answer to an action.
Catch anomalies automatically
The anomaly engine learns each metric's own baseline and its weekly rhythm, so it flags what's abnormal for a Tuesday 9am without a single hand-tuned threshold.
Plan capacity before you hit the wall
Forecast CPU, memory, disk, and interface utilization, with projected time-to-exhaustion. Disk-full ETAs and link-saturation runway arrive with confidence intervals.
Troubleshoot outages in seconds
The causal engine isolates an event's blast radius, names the likely root cause, and turns that prediction into an alert your team can act on.
Query in plain English
Describe what you want and let AI write the SRQL. Agents call the same intent-based, injection-hardened tools your operators use, with no raw string concatenation and no guessing.
Four engines that make the AI trustworthy.
The intelligence isn't a prompt wrapped around a database. It's a set of purpose-built engines (causal, topological, statistical, and predictive) that turn raw telemetry into explanations and forecasts.
Causal Reasoning Engine
A DeepCausality-powered prediction engine that reasons over live state, classifies every entity as root cause, affected, or healthy, and feeds predictions back into alerting. Automation, not just a pretty graph.
Causal Topology Engine
A GPU-rendered "God View" of your network: deck.gl and zero-copy Apache Arrow streaming, millions of nodes at 60fps, with causal overlays that light up a blast radius without recomputing the map.
Anomaly Engine
Streaming, scale-invariant detection against learned per-series baselines with day-of-week and hour-of-day seasonality. It evaluates millions of samples per second and won't let a flood poison its own baseline.
Capacity Planner
Scheduled forecasting over long-horizon rollups that fits a trend per resource, projects time-to-exhaustion with a confidence interval, and routes at-risk findings into the same alerting spine.
Open by protocol. MCP-native today.
ServiceRadar speaks the Model Context Protocol, so the assistants your team already uses can read and reason over your network through safe, intent-based tools. Our MCP server is available now, with first-class plugins and skills on the way.
ServiceRadar MCP server
AvailableConnect any MCP-compatible assistant to live, intent-based tools over SRQL. Injection-hardened and ready today.
Claude plugin
In developmentA first-class ServiceRadar experience inside Claude.
Claude & Codex skills
Coming soonPackaged skills that teach coding agents how to operate ServiceRadar.
Codex plugin
Coming soonServiceRadar integration for Codex-based developer workflows.
Put AI Ops to work on your network.
Explore the live demo, self-host the open-source platform, or talk to us about ServiceRadar Cloud and Enterprise support.