Building with AI in 2025

AI has fundamentally changed how I write code. Not replaced it - changed it. Here's what that looks like in practice.

The New Workflow

My typical workflow now involves:

1. Planning with AI - Sketching out architecture, asking for tradeoffs
2. Scaffolding - Generating boilerplate and initial structure
3. Iteration - Refining code through conversation
4. Review - Using AI to catch edge cases and bugs

What Works Well

Boilerplate elimination - The tedious parts of coding (setup, configuration, repetitive patterns) are essentially free now.

Learning acceleration - When I need to use a new library or API, I can get up to speed in minutes instead of hours.

Code review - AI catches things I miss. Not always right, but often useful.

What Doesn't Work

Complex architecture - AI can suggest patterns, but understanding your specific constraints requires context it doesn't have.

Debugging production issues - AI doesn't know your system's history, your data, your users.

Creative solutions - Sometimes you need to think outside the patterns AI has seen.

My Setup

Currently using:

• Cursor for daily coding
• Claude for longer conversations and planning
• ChatGPT for quick questions

The Key Insight

AI is a multiplier, not a replacement. It makes good developers faster, but doesn't make non-developers into developers.

The developers who thrive with AI are the ones who know when to use it and when to think for themselves.

Looking Forward

We're still early. The tools will get better. The key is building intuition for what works now while staying adaptable for what comes next.

The best time to learn these tools was a year ago. The second best time is now.