AI-Ops with Human in the Loop

AI-Ops is rising like the Olympians of Greek mythology — faster, smarter, and more collaborative than the Titans of old workflows. But even the gods needed strategy. They needed a Prometheus to bring the spark and guide their power. In AI-Ops, that’s the human in the loop: our role isn’t to write every line of code — it’s to orchestrate.

Kickstarting with Opus

Start by giving Opus a high-level description of what you need. It thinks ahead — building with debugging efficiency, and solid architecture in mind. Say if you start the project in the evening, by morning, you should have a complete, ready-to-run codebase that works.

Deep Understanding with Gemini 2.5 Pro

Use Gemini to break down the codebase. With tuned prompts, Gemini explains the architecture and logic in depth, outputting structured markdown. You can review its analysis to understand exactly what was built overnight and note any changes needed.

Precision Edits with o3

For complex changes and optimizations, turn to o3. It’s incredibly insightful — sometimes outperforming professional developers in identifying better approaches. Due to token limits, you will want to manage its output with todo list and continuation prompts.

Tool Strengths

• Opus – Best for starting new projects. It anticipates key engineering concerns (observability, debugging efficiency, etc.) and often produces working solutions on the first try.

• o3 – Extremely insightful, especially for complex changes. Sometimes its problem-solving ability surpasses that of professional developers.

• Gemini – Superior in context understanding and processing speed. Ideal for translation and code comprehension. It can systematically break down and cross-reference modules, enabling it to accurately understand even tens of millions of lines of code overnight without errors.

This AI + Human loop accelerates development, improves quality, and makes working with massive codebases not just possible — but enjoyable.