AI Disruption

AI Disruption

AI Programming Tools: Is Cloud the Endgame?

AI Coding Tools: Cloud vs. IDE vs. Terminal. Explore the future of AI programming, key capabilities, and which format leads.

Meng Li's avatar
Meng Li
Oct 02, 2025
∙ Paid
3
3
Share

“AI Disruption” Publication 7800 Subscriptions 20% Discount Offer Link.


Strictly speaking, there are 5 product formats, but they can be consolidated into 3 categories:

  1. Standalone IDE (and IDE plugins)

  • Cursor, Github Copilot, Windsurf, Cline, Augment Code

  • Wenxin Kuaima, Tongyi Lingma, Tencent CodeBuddy, Trae

  • Open-source versions: RooCode, Zed

IDE capabilities basically consist of three major components: Autocomplete - this is the most basic and high-frequency capability, existing as real-time, inline code suggestions; Chat - this is the interactive dialogue layer where developers can ask questions, explain code, and generate code snippets through natural language; Agent - this is the autonomous execution layer.

AI is no longer passively responsive but can actively plan and execute complex tasks spanning multiple files and steps, such as completing a feature development or fixing a bug.

  1. Terminal

  • Claude Code, Codex CLI, Gemini CLI

  • Qwen Code, Kode, etc.

  1. Cloud (and Github bot mode)

  • Codex, Devin, OpenHands (has an open-source version)

This post is for paid subscribers

Already a paid subscriber? Sign in
© 2025 Meng Li
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture