Nestack Agent Care
Industries / Telecom / Outage reporting agent

Telecom AI agent · outage reporting

Outage & Disaster Reporting AI Agent

Start the right Part 4 clock for the provider class this outage falls in, assemble the threshold arithmetic with the systems each figure came from, and leave the oath to a named person.

4–6 weeksTypical delivery
Your stackDeployment
Class-boundDeposed on oath
Agent CareAfter launch

What this agent does

Drafts the report, never the oath

In
01

A fault is discovered at 02:14, and every Part 4 clock runs from that moment, not from the ticket.

02

A provider class is read first — 47 CFR 4.9(g) gives interconnected VoIP no Initial Report at all.

Reason
03

A VoIP user-minute trigger also wants complete loss of service, a bar no other class is held to.

04

A 911 or 988 facility is hit, and § 4.9(h) has run a thirty-minute clock since 15 April 2025.

05

An activation is announced, and § 4.18(a) has been mandatory only since 20 February 2025.

Decide
06

A daily DIRS report is owed even when the reportable infrastructure has not changed since yesterday.

07

A waiver is read per outage — § 4.18(b) holds only if the first daily report falls due first.

Out
08

A Final Report is deposed to on oath under § 4.11 by a person authorised to bind the provider.

09

Execute write actions only inside the approval boundaries agreed during implementation.

Product statement

The agent watches and assembles; a named person authorised to bind the provider deposes on oath, files the report, and the provider answers for it.

Example workflow

One outage, discovery to filing

AgentHuman
1Discovery timestamp receivedAlarm records, a NOC ticket, element logs or a field report
2Provider class and clocks setWireline, wireless, cable, satellite or interconnected VoIP, each on the § 4.9 paragraph it answers to
3Reports draftedThe Notification, the Initial Report where a class has one, and the Final Report
4Controls appliedClock checks, waiver checks that never reach the 911 and 988 notices, and completeness confidence
No human action required

Stages 1 to 4 run unaided, and nothing is filed at any of them — the agent is drafting, and the duty lane opens at the completeness gate.

5DecisionBranches at the completeness gate
Arithmetic sound

Goes to the authorised filer to depose.

Anything thin

Adds a regulatory counsel read first.

Duty officer review

The draft is held with its clocks, the two-hour follow-up and the arithmetic behind it.

Depose and file · Append evidence · Send to counsel
Filed — by an authorised person
6Outage and case records updatedOnly where write access and § 4.2 confidentiality allow it
7Outcome evaluatedClock margin, arithmetic provenance, counsel corrections and what review found
Corrections

Each counsel correction is counted in the evaluation.

What should not run autonomously

Human approval stays in control

Outside the boundary — human approval required8 items
Deposing on oath under 47 CFR 4.11.
Filing a Notification, Initial or Final Report.
Notifying a 911 or 988 special facility direct.
Deciding the user-minute threshold was not met.
Automation boundaryAgent acts unaided
Read the discovery moment from the system that first raised it.
Start the clock that 47 CFR 4.9 sets for the provider class involved.
Assemble the user-minute arithmetic and name each source system.
Draft the Notification and the reports that provider class owes.
Nothing is filed or deposed except by an authorised person, inside agreed boundaries.
Judging an outage outside the 47 CFR 4.3 list.
Telling a PSAP that service has been restored.
Deciding a DIRS activation waives a NORS filing.
Changes to clock rules or report templates.

Example output

One outage, annotated

Our NOC incident agent names the cause; this record is the filing, and the clock ran before any cause was known.

Report draft · single outageIllustrative example
Outage
Drafted as
Clock
Evidence of record
Confidence
Held for
Transport fault, wireline metro
Notification under § 4.9(f), thresholds met
Minute 41 from discovery
Alarm and mediation records, 14 August 2026
Held unfiled
The authorised filer, in person
As receivedTaken from the alarm log and the mediation counts — the arithmetic reaches as far as those records do.
What the record holds Discovery timestamp User-minute arithmetic DIRS activation status
Why no threshold call hereThreshold judgement sits with the person who deposes under § 4.11.
ActionDepose and fileAppend evidenceSend to counsel
What the score decidesA draft under the configured threshold picks up a counsel read before the filer sees it.

Value

Where AI adds value

The same four claims, placed at the point in the workflow where each one applies.

Where the value landsValue 01 – 04
Every outageFrom the alarm that raised it
03Evidence

One discovery, two filings

Our outage communication agent drafts the customer notice from this same discovery; this page is the regulator filing, on other clocks and with another signer.

01Approved path

One event, several clocks

Wireline, wireless, cable and satellite run 120 minutes, 72 hours and 30 days; § 4.9(g) gives interconnected VoIP 240 minutes, 24 hours and no Initial Report.

02Human review

What was found, and what was not

No FCC rule reaches outage reporting for broadband internet access service, and no 2025 or 2026 enforcement for a late or missing NORS or DIRS filing was found.

04Build an evidence trail

The outage, the class it falls in and the person who deposed on oath stay on the record.

Integrations

Typical integrations

Five system groups connect to the same agent. Which of them are in scope is decided in discovery.

Filing systemsNORS · outage notifications
DIRS · daily activation reports
Network monitoringAlarms · element managers
NOC ticketing · incident feed
Traffic and subscriber recordsMediation · switch counts
Subscriber and cell inventory

Agent

Outage and disaster reporting

Reads the discovery
Drafts the reports
Holds for the filer

Public safety contacts911 special facilities
988 special facilities
Observability & evaluationOpenTelemetry · Langfuse
Supported monitoring/evaluation sources

Integration availability depends on the client's existing systems and API access.

Agent controls

Six gates between the model and the filing

Six gates, and no two of them shut on the same thing. What passes all six is named in the map below.

L6 · Outermost — last line of defenceInward → L1 · closest to the model
L6Rollback / safe modeNarrow the agent to notification assembly when evaluation or production signals degrade.Roll back
L5Version monitoringTrack model, prompt and clock-rule changes; the new § 4.18(c) is delayed indefinitely.Track
L4TraceabilityRecord the discovery, the clocks run and each read; § 4.2 presumes the report confidential.Record
L3Filer releaseHold the draft for a person authorised to bind the provider; § 4.11 wants an oath, not an officer.Gate
L2Policy guardrailsTest the draft against the § 4.9 paragraph for the class; the 2026 deletions left § 4.9 intact.Restrict
L1Confidence thresholdsDivert a thin draft into a counsel read before the duty officer sees it.Require review
Model coreReports drafted — the discovery, the clocks, the arithmetic and completeness
L1 – L2Test whether a draft may stand
L3Puts the oath in a person's hands
L4 – L5Keep the outage and the arithmetic behind it
L6Narrows to notification assembly when signals degrade

How Nestack evaluates it

Evaluate the whole assembly — not only the report that comes out.

Coverage runs the whole depth of the workflow, and every layer is cut by slice.

Surface — the report the Commission reads
Depth of coverage ▼
E1Final-output evaluationDid the draft record the moment discovery actually happened?
E2Step-level evaluationDid the agent run the clock this provider class is on, on the right counts?
E3Tool evaluationDid it read and write the correct outage record and the correct counts?
E4Confidence calibrationDo low-confidence drafts actually attract more counsel corrections?
E5Slice evaluationHow does performance change across specific provider classes?
E6Business outcomeHow many drafts needed a correction before the filer deposed?
Floor — the outcome the provider answers for

Failure modes

Where each failure originates in the agent

Seven failure modes, each pinned to the point in the run where it begins.

Agent lifecycleDirection of processing →
01 · Retrieval1 mode
IN-03

Stale discovery time

The timestamp read is not the one the alarm holds.

Stage gathersThe discovery, the class, the counts and the sources
02 · Reasoning2 modes
IN-04

One playbook, every class

The wrong § 4.9 clock is run for the class.

IN-06

Waiver read as automatic

A DIRS activation is treated as a NORS waiver.

Stage proposesThe clocks run, the arithmetic and completeness
03 · Tool / write2 modes
IN-02

Thin draft passed forward

A draft moves on without the counsel read.

IN-05

Special-facility hit missed

A 911 or 988 impact is not flagged for notice.

Stage writesOnly where write access and approval policy allow it
04 · Output1 mode
IN-01

Filed, arithmetic unrecorded

The record shows a report but not the figures behind it.

Stage returnsThe report a person deposes to and then files
05 · Change / Version1 mode
IN-07

Silent clock regression

A rule change moves a clock without moving the record.

Stage tracksModel, prompt, clock rules and report fields
Sev-1 · a report is filed without a person Sev-2 · a wrong clock reaches the record Sev-3 · source degrades, draft holds unfiled

Affected slices

One provider class absorbs the corrections

A class-level clock-adherence figure can read clean while one provider class takes most of the corrections. Nestack reports the counsel-correction rate by provider class, not only in total.

Slice performance — reported separately, not only in aggregateIllustrative example
SliceFailure rateLift Lift vs. thresholdStatus
Interconnected VoIP service9.1%3.6× Review
Cable circuit-switched telephony6.6%2.6× Review
Wireless network operations4.1%1.6× Watch
Wireline incumbent operations1.7%0.7× Normal
Bar: counsel-correction-rate lift vs. wireline-operations baseline · scale 0–4.0× · tick marks the 2.0× review threshold 2 of 4 slices over threshold

Evidence-linked improvement

What an unreconstructable filing costs

A cycle is done when the wrong provider clock has become a case the next release must pass. That suite is what the next report built is measured against.

Improvement cycle · five stagesSwitchback — the path turns at Improve and returns at Learn
01Detect

Counsel-correction rate rises in one provider class.

02Diagnose

The engineer who knew at 02:14 and the form that thought discovery was at 06:00 are read back until one cause remains.

03Improve

Number the change; the outages that drove it are filed under it.

04Verify

Each touched class case is run once more, and one red holds it back.

05Learn

It stays on as a standing test, and the threshold rules move with it.

Learn → DetectThe return edge. The next report is measured against a suite one case longer.

Typical build scope

Twelve workstreams across six weeks

The build scope read against the delivery timeline. Week structure follows the six-week plan — discovery, sources, report assembly, evaluation, integration, then production validation and handover.

Workstream Week 1Week 2Week 3Week 4Week 5Week 6
01Outage-reporting workflow discovery and boundary work.
02Alarm, ticket and mediation sources.
03Provider-class and threshold-arithmetic mapping.
04Discovery-evidence ingestion.
05Class, clock and record binding.
06Completeness scoring and counsel routing.
07Authorised filer workflow.
08Alarm and case-system integration.
09Threshold and clock cases.
10Guardrails and deposition controls.
11Outage-trail instrumentation.
12Deployment, documentation and Agent Care handover.
12 workstreams · 6 weeks · bar shows the weeks a workstream is active — several run in parallel Final scope and sequence confirmed in discovery

Engagement tiers

What each tier includes

Rows are the capabilities named in each tier's scope. Higher tiers include everything below them.

Capability✓ in scope · — not at this tier PilotOne provider class, one region ProductionProduction reporting workflow AdvancedMultiple classes / regions
Introduced at Pilot
Report drafting to your clocks
Authorised-filer deposition
Threshold-arithmetic baseline
Introduced at Production
Reporting by provider class
Filing workflow in your systems
Approved write-back
Alarm-system integration
Introduced at Advanced
Multi-class reporting duties
Cross-class evidence packs
High outage volume
Multi-clock notification controls
Build price From $5,000 From $8,000 Custom quote
Final build priceConfirmed after discovery based on integrations, workflow complexity, transaction volume, approval controls and deployment requirements.
Separate from buildBuild pricing is separate from recurring Agent Care, which covers managed monitoring, evaluations, incidents and verified improvements after launch.

What we need from you

What you bring, and what we build with it

Each input maps to a piece of build scope and a week in the delivery timeline.

You bringWe build with it
01Your network alarms and outage tickets Discovery capture and clock bindingWeek 1
02Representative outages and their counts Threshold arithmetic, source binding and the draft baselineWeek 2
03Your provider classes and filing channels Provider-class, clock and channel mapping and the automation boundaryWeek 1
04Access to relevant APIs, feeds or exports Alarm, ticket and mediation assessment, then integration setupWeek 2
05Reports you would not want deposed to Threshold cases and failure testingWeek 4
06What no outage report may conclude Completeness scoring, counsel routing, guardrails and filing controlsWeek 3
07A person authorised to bind the provider Authorised-filer workflow, then pilot and production validationWeeks 5–6
Nothing else is required Deployment, documentation and Agent Care handover are ours.

Delivery timeline

Four phases across six weeks

Each band is the honest working time a phase takes, and the fifth is shared because the work itself is.

Phase W1W2W3W4W5W6
Discovery W1
Build W2 – W3
Evaluate W4 – W5
Pilot & Launch W5 – W6
Week focus W1Outage reporting discovery, clock mapping and the automation boundary W2Source integration and the threshold-arithmetic baseline W3Report assembly, class logic and filing controls W4Evaluation suite, threshold cases and failure-mode testing W5Alarm integration, pilot outages and targeted corrections W6One activation season run under the duty officer, then Agent Care handover
Reading the bandA bar runs only across the weeks its own work is named in. The fifth week doubles because two phases share it.
At the end of W6When the outage record validates, Agent Care takes the agent on.
DurationSix-week plan shown · typical delivery 4–6 weeks depending on scope confirmed in discovery.

Next step · Telecom AI agent

Build the automated outage-report trigger a 2024 consent decree promised.

Show us how a fault reaches you today and who deposes on the Final Report. That deposition has to survive a read-back against the alarm log months later, on a clock that started at discovery. A 2024 consent decree ran on a notification filed late and maintenance never reported.

Nestack Agents · outage reportingAGT-TL-19 · Agent Care available after launch