{"id":1467,"date":"2025-11-27T06:59:21","date_gmt":"2025-11-27T06:59:21","guid":{"rendered":"https:\/\/fredborg.org\/?p=1467"},"modified":"2026-01-07T10:30:01","modified_gmt":"2026-01-07T10:30:01","slug":"why-i-dropped-github-copilot-for-roo-code-a-real-world-ai-workflow","status":"publish","type":"post","link":"https:\/\/fredborg.org\/?p=1467","title":{"rendered":"Why I Dropped GitHub Copilot for Roo Code: A Real-World AI Workflow"},"content":{"rendered":"\n

<\/h1>\n\n\n\n

If you watch AI coding tutorials on YouTube, you probably know the drill: the prompts work perfectly, the code compiles on the first try, and a complex project is finished in minutes. It looks like magic.<\/p>\n\n\n\n

But if you are a developer working on real-world architecture, you know that isn\u2019t the reality.<\/p>\n\n\n\n

I am starting a new series where I build a full-scale Business Central Webhook integration<\/strong> using AL code and the Azure stack. But I\u2019m doing something different: I\u2019m leaving the filters off.<\/strong> You are going to see the hallucinations, the logic errors, and exactly how I debug a complex ERP integration when the AI gets it wrong.<\/p>\n\n\n\n

To do this, I made a major change to my stack: I stopped using GitHub Copilot. Here is why.<\/p>\n\n\n\n

The Problem with GitHub Copilot<\/h3>\n\n\n\n

Don’t get me wrong, Copilot is a fantastic tool. But it has a hidden ceiling for power users.<\/p>\n\n\n\n

When you are doing heavy architectural AI work\u2014generating massive blocks of AL code or complex Azure configurations\u2014you inevitably hit fair usage policies<\/strong>.<\/p>\n\n\n\n

Basically, if you push it too hard, they throttle you. You run out of “fast tokens,” the responses slow down, and you get stuck waiting for a reset. In a real professional workflow, I cannot afford that bottleneck.<\/p>\n\n\n\n

The Solution: OpenRouter + Roo Code<\/h3>\n\n\n\n

My new workflow relies on two specific tools that solve the throttling issue and add massive functionality.<\/p>\n\n\n\n

1. OpenRouter (The Engine)<\/strong> Instead of a flat fee for limited “fast” access, I switched to OpenRouter<\/strong>, which uses a pay-as-you-go model. I pay for exactly what I use. If I need to burn through a million tokens to solve a hard problem, I can do that without hitting an invisible wall.<\/p>\n\n\n\n

2. Roo Code (The Driver)<\/strong> To drive OpenRouter, I am using a VS Code extension called Roo Code<\/strong>. This isn’t just a chat window; it is an autonomous coding agent<\/strong>.<\/p>\n\n\n\n

The Killer Feature: Multi-Agent Workflow<\/h3>\n\n\n\n

The real reason I made the switch to Roo Code is the multi-agent workflow<\/strong>. It\u2019s like having a full dev team inside VS Code, where every member is specialized for their specific job.<\/p>\n\n\n\n

The true power here is that each agent can run its own specific LLM.<\/strong> You aren’t locked into one model for everything; you can pick the best tool for the task:<\/p>\n\n\n\n

    \n
  • The Orchestrator:<\/strong> This agent acts as the Product Manager, breaking down plans. For this, we can use Gemini<\/strong>, leveraging its massive context window to keep the entire project scope in view.<\/li>\n\n\n\n
  • Code & Debug Agents:<\/strong> These handle the actual syntax and fixing errors. Here, we can swap to Claude<\/strong>, because it is incredibly fast and produces high-quality code.<\/li>\n\n\n\n
  • Documentation Writer:<\/strong> When it’s time to write docs, we can assign a custom agent powered by ChatGPT<\/strong>, which excels at natural language and explanations.<\/li>\n<\/ul>\n\n\n\n

    This flexibility means we are never compromising. We get the best context, the best code, and the best writing, all working together in one orchestrator-led system.<\/p>\n\n\n\n

    Join the “Unfiltered” Journey<\/h3>\n\n\n\n

    This series is about learning how to actually<\/em> work with AI, not just showcasing a magic trick.<\/p>\n\n\n\n

    I will be recording my screen and sharing the workflow as it happens. If the AI hallucinates a table that doesn’t exist in BC, you\u2019ll see it. If Azure rejects our deployment, we will fix it together.<\/p>\n\n\n\n

    If you want to see if AI can actually handle a real-world Business Central project\u2014scars and all\u2014subscribe to the channel<\/strong> and follow along.<\/p>\n\n\n\n

    \n