Open Source Telco-Specific AI Agents and Skills

Problem Statement

The telecommunications industry faces a significant gap when it comes to leveraging AI effectively. General-purpose AI models and tools lack the domain-specific knowledge required to understand telco architectures, protocols, and operational workflows. This results in CSPs either building proprietary solutions from scratch - duplicating effort across the industry - or settling for generic tools that require extensive prompt engineering and manual adaptation to be useful in a telco context.

Description

This project brings together engineers from multiple CSPs to collaboratively identify, develop, and open-source AI agents and skills tailored to telecommunications use cases. Participants should ideally come from similar domains (e.g., Core Network, Cloud Native Automation, RAN, Transport) so that concepts and skills can be validated cross-company and ensure real-world applicability.

The focus is on creating reusable, modular AI skills that encode telco-specific knowledge - from network configuration patterns to troubleshooting procedures - and packaging them as open-source artifacts that any operator can adopt and extend.

Why Open Source?

By open-sourcing these agents and skills, the telco community benefits in several ways:

  • No need to reinvent the wheel -- build once, share everywhere, and avoid duplicated effort across operators
  • Community-driven improvement -- skills get refined and hardened through contributions from multiple organizations with diverse environments
  • Accessibility for smaller operators -- telcos with limited AI/ML teams gain access to powerful, production-ready tools they couldn't build alone
  • Vendor and operator interoperability -- shared skills create a common language and approach to AI-assisted network operations
  • Faster time to value -- pre-built skills reduce the ramp-up time for teams adopting AI in their workflows
  • Transparency and trust -- open-source code can be audited, ensuring no hidden biases or vendor lock-in in AI-driven decisions

Example Use Cases

  • Automated troubleshooting agents for common network faults (e.g., signaling failures, capacity degradation)
  • Configuration generation and validation skills for vendor-specific network functions
  • Intelligent alarm correlation and root cause analysis
  • Network planning and capacity forecasting assistants
  • GitOps workflow agents that understand telco deployment patterns and dependencies

Project Details

Leader: Swisscom (J. Weber, J. Studler)

List of people/organizations interested to join:

Hackathon Objectives

  • Identify telco-specific AI skills that can be open-sourced and shared across companies
  • Define agent use cases applicable to multiple operators' networks
  • Evaluate existing models and initiatives such as the Large Telecom Model (LTM) or GSMA's Open Telco AI Initiative
  • Prototype at least one reusable skill or agent during the hackathon
  • Stretch: Establish a governance model and repository structure for ongoing community contributions