Pioneering AI Agents to Automate Enterprise Workflows

Automating complex network operations for a global internet leader with a code-generating AI agent

DOWNLOAD PDF

Success!

Oops! Something went wrong while submitting the form.

Codetribe partnered with AMS-IX to automate complex hardware provisioning with a sophisticated AI agent. We developed a powerful code-generating agent powered by LLM, which has domain knowledge of more than 11,000 hardware scripts, converting simple user prompts into precise command scripts. This transformed a high-expertise manual process into a streamlined workflow, setting a new standard for operational efficiency.

Client

AMS-IX

Country

Netherlands

Market

Globally

Type

Enterprise

Story

A vital, neutral hub for the internet for over 25 years, Amsterdam-based AMS-IX connects thousands of networks, enabling the seamless exchange of global data traffic. To maintain their position as an industry leader and future-proof their critical infrastructure, they partnered with Codetribe to pioneer the use of AI agents for automating complex network operations.

Business Impact

Driving Operational Cost-Efficiency

The solution provides a 20% increase in overall team capacity and allows for 50% of engineering resources to be reallocated from repetitive maintenance to high-value innovation. By automating routine operations and cutting support costs, the agent empowers key personnel and allows AMS-IX to scale its capacity and reliability without increasing headcount, directly boosting cost-efficiency.

Drastic Improvement in Quality and Reliability

By automating the complex and manual task of script generation, the AI agent is projected to cause a 55% reduction in human-induced errors. This significantly boosts network reliability, reduces slow-downs from manual mistakes, and streamlines incident response, ensuring a more stable and resilient service for their global partners.

Laying the Foundation for Future AI

This project successfully proved the business case for AI, creating the foundation for a future AI-driven user experience for both customers and internal engineers. By validating this agent-based approach, AMS-IX can now move forward with plans to delete 75% of its legacy script library, paving the way for a more efficient, automated, and future-proof technology stack.

55%

Reduction in Human-Induced Errors

50%

Reallocation of Engineering Teams to Innovation

20%

Increase in Overall Team Capacity

75%

of Legacy Scripts Targeted for Deletion

Focus Area

Domains

AI & ML Solutions

Conversational AI

Enterprise Software

Web Apps

API

Architecture

Hardware

Services

AI & ML Services

Digital Transformation

Infrastructure & DevOps

Back-End Development

Front-End Development

Prototyping & POCs

System Architecture

Web Development

Industries

Artificial Intelligence

Telecommunications

Tech Stack

Python

RabbitMQ

React.js

Unsupervised Learning

Reinforcement Learning

Large Language Models (LLMs)

Retrieval-Augmented Generation (RAG)

Prompt Engineering

Challenges

AMS-IX's network operations, critical to global internet traffic, were constrained by legacy processes. Growth was stalled by siloed knowledge and operational bottlenecks, with key engineers stuck on maintenance instead of innovation. The core challenge was a library of over 11,000 complex scripts, which required slow, manual interpretation for any provisioning, leading to a high risk of human error and slow incident response. Modernizing this system required a new, advanced solution that was fail-safe, seamlessly integrated, and capable of empowering the team.

Solutions

Codetribe delivered an end-to-end AI agent solution by first training the agent on the entire library of 11,000 scripts, allowing it to learn the complex patterns and meaning behind their network operations. We then adapted and fine-tuned a world-class LLM, creating a bespoke code-generating agent (dubbed the "Code Genius") that could interpret simple user prompts to autonomously generate the exact API calls and command scripts needed. To ensure reliability, we engineered a robust command messaging bus, creating a human-in-the-loop for manual approval, and validated the entire approach by successfully integrating the agent with the client's APIs in a controlled environment.