Can an AI agent be a network engineer—reset a Cisco router password

When discussing replacing IT professionals with AI, network administrators smile and say: "This isn't about my profession, much must be done by hand." But what if AI tried it on a real task?

Task Selection

Generating configs and diagnosing logs is too simple to verify. A more complex task was chosen: reset a forgotten password on a Cisco router. This requires:

• Connect via COM port (serial console)
• Reboot the router
• Catch the 15-second window and press interrupt (Ctrl+Break)
• Enter ROMMON (recovery mode)
• Load the configuration into memory
• Reset the password in the config file
• Save and reboot

This requires knowledge of specific timings, understanding of router states, work with the file system.

Result

The AI agent succeeded. The model understood the sequence, managed the COM port, tracked router output, adapted to unexpected responses. The agent didn't just follow instructions—it reasoned about why certain steps were needed and adapted to real equipment.

Although there were moments when the model got confused about command syntax or misinterpreted console output, overall the agent showed it can work with the "black magic" of networks.

What This Means

AI won't replace a network engineer 100%, but it can tackle routine, predetermined tasks. Resetting a password is one task out of hundreds. But if AI can automate it, it can also diagnose configuration issues, plan migrations, write scripts for bulk operations.

The question isn't whether AI will replace the network engineer. The question is which routine operations can be delegated to an agent so the engineer can focus on design and optimization.