du and Comarch Communications Launch Agentic AI for OSS Network Operations

du, the leading telecommunication and digital service provider and Comarch Communications, a global provider of Enterprise Tech solutions for connectivity, have expanded their partnership to lay the data foundation for goal-oriented autonomous execution. Comarch Communications is delivering a major upgrade to du’s Operational Support Systems (OSS) platform alongside specialized Inventory and Topology Management (ITM) enhancements.

Modern telecommunications networks require highly accurate, real-time data models to maintain full visibility across multi-vendor physical, logical, and virtual resources. By establishing a unified, vendor-agnostic "single source of truth" within Comarch Communications’ OSS suite, the platform ensures that network topology and resource mappings are continuously updated. This data foundation enables goal-oriented Agentic AI to perceive network conditions, analyze multi-vendor logs, and execute complex workflows without processing errors.

Justin Shields, Chief Information Officer at du, said “Our ambition at du is to raise the standard for operational agility while keeping trust of our customers, resilience and control is at the center of how we plan to scale AI. This next phase with Comarch Communications strengthens the inventory and topology foundation that underpins our vision on enhanced customer experience, while enabling a new generation of AI-enable applications and workflows. By combining trusted network intelligence, embedded AI and automation, we are creating the foundation for increasingly autonomous operations and faster decision-making. This is an important step towards simpler, more proactive and more intelligent operations across our multi-brand environment.”

Paweł Workiewicz, Chief Commercial Officer at Comarch Communications has added: "Seven years after the beginning of our work together, we are taking our collaboration with du to the next level. With network maintenance and data reconciliation happening automatically in the background thanks to Agentic AI, subscribers experience direct benefits: higher reliability and fewer outages. This platform upgrade directly improves network uptime and ensures faster troubleshooting across all digital services."

Transforming Operations with Agentic AI

The collaborative initiative introduces goal-oriented, autonomous AI agents capable of perceiving network conditions, analyzing complex multi-vendor system logs, and executing automated workflows. Key operational capabilities include:

• Agent-Assisted Root Object Missing-Data Detection: Identifying missing or inconsistent configuration attributes across vendor Network Management Systems (NMS) and autonomously executing data restoration, synchronization, and re-discovery jobs without human intervention.

• AI Agent-Driven Sub-Component Relationship Analysis: Utilizing specialized Large Language Models (LLMs) and hierarchical AI agent architecture to instantly diagnose structural network discrepancies, distinguish between vendor export errors and internal configuration gaps, and automatically deploy corrective updates directly to inventory.

• Agentic AI: Automated Remediation and Smart Ticketing: Lowering operational Mean Time to Resolution (MTTR) from several days to minutes by unburdening engineering teams from manual database text-searches and raw XML analysis, automatically restoring errors where feasible, and generating smart tickets for complex issues.

• AI-Driven Resource Catalog Modeling and Mapping: Eliminating manual cataloguing by allowing AI agents to autonomously read vendor technical documentation, extract specifications, and instantly publish complex hardware object hierarchies and position mappings in the Device Library via automated APIs.

To ensure enterprise-grade security and operational safety, the AI framework operates within strict governance guardrails. While initial deployment utilizes secure "sandbox" environments with human-in-the-loop verification, the architecture is designed for goal-oriented autonomous, closed-loop execution. 

As the AI models mature, these agents will independently validate, approve, and push network planning adjustments to the live production network, achieving true zero-touch inventory operations. The framework remains model-agnostic and designed for secure on-premise deployment.