Agentic AI-Powered Network Operations & Customer Experience Platform
CLIENT
Proline
YEAR
2026
Overview
The provider's operational load had grown by 340% subscriber growth in 36 months, which overwhelmed its existing manual and reactive customer support systems.
To address this, Marlocks Technologies (AWS Partner) built an Agentic AI-powered Network Operations and Customer Experience Platform on AWS.
The platform automates network monitoring, fault resolution, fraud detection, and customer support using AI-based classification services. It improved network reliability, reduced operational workloads and significantly enhanced customer experience while meeting NDPR and PCI DSS v4.0 compliance requirements.
Key Challenges
Reactive Network Fault Management
Network anomalies were mostly detected through customer complaints, leading to slow response times and heavy operational loads.
Billing Fraud & Revenue Leakage
Fraud activities such as credential sharing and subscription misuse caused an estimated ₦30M monthly revenue loss.
Fragmented Network Monitoring
Traps of network management systems lacked a unified observability layer, making root-cause analysis difficult.
Low Customer Support Resolution
Customer support handled 18,000+ monthly interactions, but the first-call resolution rate was only 41%, causing low customer satisfaction.
The Solution
Operational Efficiency
Marlocks Technologies implemented a four-layer AI-driven architecture on AWS:
AI Intelligence Layer
Two AI agents built with Amazon Bedrock:
NOC Assistant - Detects, analyzes, and resolves network faults automatically
Customer Service Assistant - Handles customer queries, troubleshooting, and service requests
Orchestration Layer
AWS Step Functions automate workflows such as:
Network fault detection and remediation
ITSM escalation and SLA monitoring
Automated subscriber provisioning
Data & Integration Layer
Amazon OpenSearch - unified network monitoring and observability
Amazon DynamoDB - subscriber data and fault ticket storage
Amazon EventBridge - real-time event integration across systems
Orchestration Layer
The platform ensures compliance with:
Nigeria Data Protection Act (NDPR)
PCI DSS v4.0
Security tools include AWS WAF, CloudTrail, GuardDuty, and Macie
Production Results (First 90 Days)
| Metric | Before Deployment | After Deployment |
|---|---|---|
| Network Fault Resolution (MTTR) | 4.3 hours | < 37 minutes |
| Fault Detection (MTTD) | 47 minutes | 6 minutes |
| Auto Fault Resolution | 0% | 68% of Tier 1-2 faults |
| First Call Resolution | 41% | 78% |
| Average Handle Time | 11.3 minutes | 5.1 minutes |
| Customer Satisfaction (CSAT) | 56% | 81% |
| Fraud Detection Accuracy | 58% | 91.4% |
| Subscriber Provisioning | 48 hours | < 4 hours |
| Platform Availability | 97.3% | 99.91% |
| Monthly Revenue Leakage | ₦32M | ₦3.8M |
Key Business Impact
68% of network faults resolved automatically
Customer support efficiency doubled
₦28.4M monthly revenue recovered from fraud detection
Provisioning time reduced from 48 hours to under 4 hours
Full NDPR compliance achieved
Key Lessons Learned
AI tool permissions must align with network change management zones:
AI remediation actions should be restricted by infrastructure layers.
Unified data schema is critical for observability:
A standardized event schema significantly improves fault correlation accuracy.
Fraud models must be retrained continuously:
Fraud detection models need regular updates as attack patterns evolve.
AI systems must follow strict data minimization rules:
Only essential subscriber data should be used in AI model contexts to maintain compliance.
Let's Discuss Your Needs
For enquiries related to our services, media opportunities, or
partnerships, please contact our team using the form below.