After decades of coding and building AI solutions, I learned that picking the right LLM isn’t about finding “the best one” – it’s about matching each tool to the right job and knowing how to use them properly.
The story
After decades of writing code and nearly a decade building AI solutions, I just figured out something that changed how I work: picking the right LLM isn’t about finding “the best one.” It’s about knowing which tool solves which problem.
I used to waste time asking ChatGPT to autocomplete my code or trying to get Copilot to explain architecture. Wrong tools for the job.
Here’s what works:
Copilot is unbeatable for inline suggestions and boilerplate. When I’m in VS Code and need to fill out standard patterns fast, nothing else comes close.
ChatGPT is for when I need to think out loud. Architectural decisions, debugging explanations, documentation drafts – it’s conversational and helps me work through complex tradeoffs.
Claude handles the heavy lifting. When I need to analyze an entire repository without losing context, Claude is the only one that can keep up.
But here’s the thing: knowing which model to use is only half the battle.
The real skill is in how you use them. Prompt engineering, modularization, testing, iteration – that’s what separates people who get value from AI and people who just generate noise.
I break problems into chunks. I write specific prompts. I review everything. I test constantly.
The future isn’t about generating the most code. It’s about orchestrating the right tools to build something that actually works.
Recent Comments