Understanding Network CLI Configurations with ML, ANN, and LLMs

Course Q&A Generative AI with Large Language Models
1

Hi everyone,

I’m exploring how machine learning (ML), artificial neural networks (ANN), and large language models (LLMs) handle network configurations, specifically those from Cisco and other vendors using command-line interfaces (CLI).

Has anyone worked on projects where these models successfully parse, analyze, or automate network configurations? How effective are they in understanding vendor-specific CLI structures and applying network automation?

I’d love to hear insights on best practices, challenges, and any pre-trained models or fine-tuning techniques that improve accuracy in this domain.

Thanks in advance for your thoughts!

1 Like
2

Dear @rafacomu,

Welcome to the Community!

Yes, this is absolutely possible, and LLMs are increasingly effective in automating and analyzing network configurations from CLI-based systems like Cisco, Juniper, and others.

:white_check_mark: Solution Summary:

  1. Parsing CLI Output:
    LLMs can convert unstructured CLI outputs (e.g., show run) into structured formats (JSON/YAML) using prompt engineering or fine-tuning on vendor-specific data.
  2. Intent to CLI Translation:
    Models can understand user intent (e.g., “enable OSPF”) and generate corresponding CLI commands using few-shot prompting or instruction tuning.
  3. Automation Integration:
    Combine model output with tools like Ansible, Netmiko, or Nornir for execution, validation, and change control.

:wrench: Best Practices:

  • Preprocess with regex or simple parsers
  • Fine-tune with real config samples
  • Validate all outputs before deployment

:warning: Challenges:

  • Vendor-specific syntax variations
  • Ambiguity in input without strict prompts
  • Limited public config datasets

LLMs show great promise when paired with traditional network tools, especially in hybrid AI + rule-based automation workflows. Happy to share more if you’re exploring implementation.