Autonomous networks, fewer outages: How Swisscom and Outshift are rethinking network change management

Even the world’s most advanced telecom networks can be brought to a standstill by a single misconfiguration. 

According to Gartner, 80% of mission-critical outages would stem from people and process issues—with over half tied to change and configuration missteps. Globally, Splunk and Oxford Economics estimate that downtime costs organizations over $400B annually in lost productivity, revenue, and recovery efforts. The stakes for operational resilience have never been higher. 

These challenges impact industry leaders like, Swisscom, Switzerland’s largest telecom provider. When Swisscom experienced a significant incident, it became the catalyst to rethink network reliability.  

Earlier this year, Swisscom and Outshift by Cisco announced a strategic partnership to explore autonomous network operations. Today, we're proud to share the results of that collaboration. 

Our shared goal was clear: shift network operations from reactive to predictive. By combining AI agents, knowledge graphs, and network digital twins (NDTs), we demonstrated that a real-world outage can be proactively prevented, showcasing a smarter, more resilient future for telecom infrastructure.  

The vision: Toward a flawless, autonomous network 

Swisscom has long been a pioneer in telecom innovation, launching one of Europe’s first nationwide 5G networks. Their next goal is to move from manual, human-centric operations to an AI-assisted framework, ultimately aiming for full autonomy. 

But modern networks are intricate. Even the most rigorous physical testing can miss subtle dependencies, leading to service disruptions. Swisscom needed an approach that could predict and prevent failures before they reached production. 

The catalyst: Learning from a real-world outage 

Our partnership was born from a service outage triggered by a routine configuration change. Though the change passed all standard lab tests, it caused a disruption in production. 

Post-mortem analysis revealed two root causes: 

1. Incomplete test coverage: The lab missed a second-order effect on a dependent service. 
2. Misconfiguration: A missing route metric created a forwarding loop. 

This outage exposed the limitations of physical labs: slow, costly, and unable to fully replicate live environments. Swisscom teamed up with Outshift, as we were incubating an Agentic Network Validator, to develop a predictive, scalable solution. 

"At first, I was skeptical about achieving tangible results due to the project's tight timeline and ambitious scope. However, my doubts were quickly dispelled. Outshift’s Agentic Network Validator solution demonstrated its remarkable potential by analyzing real-world network changes, enabling it to predict and prevent errors before they occurred. This technology will change the way we operate networks, and the transformation will happen much more swiftly than I ever anticipated." -Martin Gysi, System Architect at Swisscom 

The proof of concept: Turning hindsight into foresight 

Together with Swisscom, we recreated the outage scenario to demonstrate how it could be proactively prevented. Our shared objectives: 

• Build a high-fidelity network digital twin: Using real topology, state, and dependency data. 
• Detect failures before deployment: Simulate changes and identify risks in advance. 
• Demonstrate multi-agent collaboration: Use Outshift’s Agentic Network Validator natural language interface to interact with specialized AI agents. 

The workflow: How AI agents prevented an outage 

1. Request: The change request was submitted in the ITSM tool using natural language. 
2. Analysis: The assistant agent interpreted the request and linked it to the official ITSM ticket. 
3. Prediction: The impact agent analyzed the network digital map to predict the full blast radius—including hidden dependencies. 
4. Testing: The test planner agent designed targeted tests based on identified risks. 
5. Validation: The test executor agent applied the change to a forked snapshot of the network digital twin, running formal verification tests.  

The result? A clear failure signal, flagging the exact misconfiguration that caused the original outage. After correcting the change, all tests passed, and the outage was avoided. 

Business impact: From technical wins to tangible value 

Deploying multi-agent validation led to transformative outcomes: 

• ✅ Fewer outages: Proactively catching failures before they reach production protects revenue and customer trust. 
• ⚡ Faster rollouts: Automated validation streamlines deployments, eliminating slow, manual lab cycles. 
• 💰 Lower costs: Significant savings from reduced manual effort and fewer costly, unplanned disruptions. 

The road to autonomy starts here 

Our collaboration with Swisscom is more than a successful proof of concept—it’s a blueprint for the future of network operations. By combining AI agents, network digital twins, and knowledge graphs, we’ve shown that autonomous networks are not just aspirational—they’re achievable. 

Building on this momentum, Swisscom and Outshift are continuing to explore how elements of the solution could be integrated into future network operations. This ongoing collaboration reflects a shared commitment to advancing predictive, agentic AI approaches that strengthen network reliability and resilience. 

Want the full story? Check out our whitepaper to learn more about the architecture and agent workflows and see how Swisscom and Outshift tackle network reliability head-on.