Live Operations — Real-time AI Alert Intelligence
Sempra Digital Brain · Live pipeline active
0
Total Alerts
Received from source today
0
Processed Alerts
Analyzed by AgentCore
0
Duplicate Alerts
Already seen similar alert
Alerts processed today
0
↑ 12% vs yesterday
Issues caught before impact
0
↑ Avg — min early warning · view details ›
Mean time to detect
0s
↓ 94% vs manual triage
SLA breaches prevented
0
ops acted on early warnings
Systems at risk — Early Warning Details
✕ close
No early warning alerts in current view.
📡 Live Alert Stream
View all alerts →
🗂️ Alert Sources
MyAccount
Customer portal
0
Active
C4C
CRM / SAP
0
Active
Layer 7
API gateway
0
Active
EDIX
EDI exchange
0
Active
ServiceNow
ITSM tickets
0
Active
AI Confidence by Domain
MyAccount patterns 96%
C4C / SAP correlation 88%
Layer 7 routing 91%
EDIX batch logic 82%
🤖 Agent Actions Taken
SummaryPer-Alert Decisions
🔮 Early Warning — Issues Caught Proactively
📅
38
min avg
Before customer impact
AI pattern match detects degradation trend before threshold breach
⏰
0
this month
Pre-emptive alerts raised
Flagged before SLA breach across all monitored applications
✅
0%
acted on
Teams resolved before incident declared
Engineering acts on AI early-warning — no SLA breach, no ticket
📊 Application Performance
| Application | Alerts | Auto-resolved | Open | P1/P2 |
|---|
🎯 Severity Split
Intelligent CIS Health Analyzer
CISQ · Beta · SDG&E · A Sempra Energy Utility
CIS
360
Last refreshed
--:--:--
Overall CIS Health Score
HEALTHY96.49%
Technology
95.72%
Business
97.26%
● Technology
Operation
98.0%
Batch 98% · SAP 100%
Availability
97.1%
Cloud 96% · SAP 99%
Integration
91.6%
Messages 96%
● Business
KPI
99.2%
Billing 100%
Transactions
94.8%
Backoffice 96%
Exceptions
98.3%
Billing 98%
📡 Source Health
Live status of each alert source feeding the Brain.
Health Insights
✓
SAP Cloud systems running OK
SAP Uptime 100%, Availability 99%, Cloud 96%. All components within normal parameters.
i
4 KPIs below threshold · batch completed
Four KPIs under threshold but batch processing completed on time. Incidents 86%, Connected 90%.
!
High priority alert — middleware
Integration score impacted — SAP integration 87%, Alerts pipeline 85%.
$
123 new billing exceptions · 345 closed
Net reduction of 222 open exceptions. Billing exception rate trending favourably.
Alert Processing Flow
How ~95,000 raw daily alerts are distilled through 6 intelligence stages into a handful of focused, actionable signals.
1 · Ingest2 · Deduplicate3 · Correlate4 · Classify5 · Enrich6 · Route
Reduction
95,000 → 112 actions (99.9% noise reduction)
Stage 1–2 · Ingest & Dedupe
95,000 → 58,900
Kinesis & EventBridge ingest from all sources. Identical alerts within 60s from same host collapse into one signal.
Stage 3 · AI Correlation
58,900 → 1,485
Bedrock multi-agent reasoning clusters related alerts across apps sharing a root cause into one group.
Stage 4 · Probable Incidents
1,485 → 460
AI scores each group by severity, system, and business impact. Known-harmless transients suppressed.
Stage 5 · Enrichment
460 → 249
Bedrock RAG matches SOPs, runbooks & architecture docs, attaching a recommended resolution path.
Stage 6a · Escalate
2 incidents
Only the most critical reach on-call — with full root-cause context, system map, and SOP already attached.
Stage 6b · Tickets & Warnings
50 tickets · 60 warnings
Enriched ServiceNow tickets for medium issues. Early warnings sent before any SLA threshold is breached.
KPI Impact Dashboard
How the Digital Brain directly protects contractual KPIs across customer experience, availability, billing, and operations.
Filter:
0
High-impact KPIs
0
Medium-impact KPIs
0
Total KPIs tracked
100%
Brain coverage
Business Value & Outcomes
Measurable impact across MyAccount reliability, C4C field operations, Layer 7 API availability, and EDIX transaction integrity.
71%
Alert noise reduction
AI correlation collapses cascading alerts into single enriched signals — engineers see the problem, not the noise.
38min
Average early warning lead time
AI detects emerging patterns an average of 38 minutes before an SLA threshold is breached.
$1.5M
Estimated annual savings
Engineering hours recaptured from manual triage plus SLA penalties avoided through proactive detection.
⚡
MyAccount — always-on experience
Login, billing, and usage alerts auto-resolved before customers notice. Session anomalies detected 40 min ahead.
💼
C4C — field crew continuity
CRM sync failures and work-order dispatch errors auto-triaged, reducing field crew downtime.
🔭
Proactive risk prevention
Brain identifies degradation trends 38 min before breach. Teams get enriched alerts with SOP recommendations.
📦
EDIX — transaction integrity
EDI 810/850/856 delays detected in real-time. Auto-retry and partner notification prevents revenue leakage.
CIO Executive Summary
Sempra Digital Brain unifies MyAccount · C4C · Layer 7 · EDIX · ServiceNow into a single AI-powered early warning system — turning 100K+ daily alerts into proactive intelligence, preventing SLA breaches and recapturing $1.5M/year.
Settings & Access Control
Configure dashboard visibility and manage user roles & privileges.
🗂️ Dashboard Tab Visibility
Choose which dashboards are visible. Hidden tabs are removed from the navigation for everyone.
🛡️ Roles & Permissions
Privilege levels available in the platform.
👥 User Management
Assign roles and manage access for your team.
| User | Role | Status |
|---|
Reports & Exports
Generate operational reports and export live data. Executive dashboards are published in Amazon QuickSight.
⚡ Quick Export
Download the current live view — respects any active application filter.
📊 Scheduled Reports
📅
Daily Incident Summary
Overnight batch + incident recap for the 6 AM shift handover. Auto-emailed to on-call leads.
Daily · 06:00 PT
🛡️
Weekly SLA Breach Report
Every SLA breach & near-miss with root cause and early-warning lead time. For service managers.
Weekly · Mon 08:00 PT
📈
Monthly Executive Review
6-sheet leadership pack — availability, value & ROI, agent performance. Published in QuickSight.
Monthly · 1st
ℹ️ Executive & trend dashboards live in Amazon QuickSight (Aurora-backed, SPICE refresh). This screen covers real-time operational exports.
Help & Documentation
Understand the metrics, the pipeline, and who to contact.
🧭 How the pipeline works
- Ingest — Source alerts (Control-M, AppDynamics, Splunk) arrive via a single mailbox and are captured through Amazon SES.
- Reason — AWS Bedrock AgentCore deduplicates, correlates, scores severity, and enriches with runbook context.
- Write — Enriched alerts land in Aurora + OpenSearch and stream to this dashboard live over WebSocket.
- Act — The Brain routes, raises tickets, escalates, or suppresses — surfacing only what needs a human.
📖 Metric glossary
- MTTD
- Mean Time To Detect — average time from an issue starting to the Brain flagging it.
- Early warning
- An alert raised before an SLA threshold is breached, giving teams lead time to act.
- SLA breach prevented
- An early warning a team acted on, avoiding a contractual breach.
- AI confidence
- How certain the model is in its classification/routing for a domain (0–100%).
- Severity
- CRITICAL / HIGH / MEDIUM / LOW — the business impact of the alert.
- Agent action
- What the Brain did: routed, ticket created, escalated, or suppressed.
📚 Runbooks & escalation
🆘 Support
On-call (P1/P2)PagerDuty · #digital-brain-oncall
Platform teamdigital-brain@socalgas.com
Dashboard issues#sempra-digital-brain (Slack)