Using conversational AI as the new IDE

For many tasks, my main development environment is now a conversation with Google's Gemini in Google AI Studio. That reduces the need for a traditional code editor and AI plugins.

From augmented typing to conversational flow

AI coding assistants started with autocompletion and snippets inside IDEs such as VS Code. Those tools boosted productivity. My own Fluid Dev process now runs mostly as a conversation, in four steps:

1. Start with a conversation. I explain my thinking, give the AI the context for the problem and describe the outcome I want.
2. The AI generates code step by step. It does not just give snippets. It guides me through the process and generates near-perfect code one step at a time. It works like an expert pair programmer who anticipates my needs and structures the solution.
3. Copy, paste and run. The generated code is often complete enough that I copy it from AI Studio, paste it into the execution environment and run it.
4. Refine with feedback. If something breaks or needs to change, I send the AI screenshots of errors, stack traces or a description of the unexpected behavior. That context lets it give targeted fixes or alternative code.

The diminishing role of the code editor

Fluid Dev relies less on AI-powered code editors for core logic. The heavy lifting happens in the conversation. The IDE is often where I run and debug code rather than where I create it. Many common tasks no longer need line-by-line typing or heavy use of in-editor AI suggestions.

Why this shift matters

• It lowers the barrier to entry. Complex coding tasks become easier to approach, because the AI guides the process.
• Prototypes come faster. Ideas become working prototypes quickly, and the back-and-forth with the AI makes it easy to try different approaches.
• The focus moves from how to what. Developers can concentrate on the problem and the functionality, not on syntax or boilerplate.
• The AI is a collaborator. It offers guidance, generates solutions and helps troubleshoot, so it is an active participant rather than only a tool.

Human oversight still matters

AI-generated code still needs human oversight, testing and an understanding of the underlying principles. For many developers, the chat window is becoming the main place where ideas are discussed, refined and turned into working code.