Xect AI is the agentic layer between the NOC and the customer. When alarms carry topology and customer mapping, agents correlate them to a likely root cause, map the affected addresses and open tickets, and drive care context and customer actions as impact is assessed, with the NOC accountable for the incident and agents supporting the customer-facing response.
An alarm carries a topology ID and a severity, nothing about who it affects, so the network gets fixed while the customer finds out by calling. When topology and customer records are mapped, Xect AI agents turn that same alarm into a scoped incident with its own list of affected addresses and a customer-facing response, as soon as correlation produces a likely cause with a confidence score attached.
Element managers, SNMP traps and telemetry streams land as events with topology IDs. Nothing about an alarm gets typed in by hand.
Alarms dedupe to one root cause using the topology already mapped in the knowledge graph. A confidence score travels with every correlation the agent makes.
Root cause becomes affected addresses and customers through Location Intelligence and the Customer 360. SLA commitments and priority accounts get checked against that list in the same step.
Customer actions, care context, the NOC ticket and restoration checks all run from one flow. Credits compute from measured downtime once the incident closes.
Each agent owns one step in the alarm-to-credit path above and hands off cleanly to the next.
Turns an alarm storm into a correlated incident and likely cause
Dedupes the flood of alarms a single fault throws off down to the fault itself, using the network topology and the knowledge graph, and holds a confidence score against every call it makes.
Turns a correlated root cause into a customer list
Takes the correlated root cause and maps it to the addresses and customers it actually affects, checking that list against SLA commitments and flagging any priority account inside it.
Triggers the customer-facing response, hands off to care and dispatch
Triggers the customer-actions flow the moment impact is confirmed, updates the care banner, IVR and chatbot together, and keeps the NOC ticket and restoration checks moving in the same flow. Notifications, dispatch coordination and credit issuance run in the Outage Customer Actions Agent flow this triggers.
Alarm, incident, impact and customer action run in one flow instead of a chain of manual handoffs between the NOC and care. This measures the incident and response workflow, not the time to physically repair a fault such as a fiber cut.
Playbooks run the moment the incident opens, without waiting on a ticket to reach the front of a queue. Auto-resolution covers remote and software remediation; a physical fault such as a fiber cut still needs a field repair.
Care, IVR and chatbot update automatically the moment the incident opens, with no separate data pull.
Every action, from the first alarm to the credit issued, traces back to a single record, so audits and disputes have one source of truth.
The NOC stays accountable for the incident. Agents support the customer response.
When an OLT alarm hits a PON, the agent correlates it to the fiber segment at fault and maps every ONT downstream to its address.
See the industry pageWhen a node alarm fires on the CMTS, the agent maps the affected modems and updates care before call volume climbs.
Talk to us about cable and HFCWhen a RAN or core alarm degrades a cell, the agent maps the subscribers on it and drives the customer actions for that footprint.
Talk to us about mobile networksWhen a SCADA alarm trips a feeder, the agent maps the meters on that feeder and updates the customer-facing outage status.
Talk to us about utilitiesWhen a facility alarm hits a data center, the agent maps the dependent services and customers and opens the incident before support tickets pile up.
Talk to us about data centersOwns the alarm queue and the time to restore, and needs every alarm to carry its business impact, not just its topology ID.
Owns the topology and the root-cause diagnosis, and needs alarms deduplicated before they ever reach an engineer's queue.
Owns the customer-facing response to an outage, and needs the care banner, IVR and chatbot updated the moment the NOC sees the alarm, not after the first call.
Owns the connectors between the NOC and everything downstream, and configures element-manager, SNMP trap and telemetry connectors, plus topology and customer mapping, so alarms land as governed events.
Book a 30-minute demo and see how an alarm becomes a correlated incident, an impact list and a care banner on your topology.