I tried every AI code editor so you don't have to

Your Average Tech Bro

132 views • 10 months ago Save 8 min 7 min read

Video Summary

AI code editors largely share underlying LLMs, making their performance similar. The key differentiators are user experience, interface, and feature set relative to cost. Dedicated AI IDEs offer integrated UIs with features like accepting individual code changes and tagging files, but can be more expensive and offer less model usage compared to direct subscriptions due to intermediary costs.

Plugins for existing IDEs, while functional, are seen as a watered-down version of dedicated IDEs, with limitations in UI and presentation. CLI tools, often from model creators themselves, provide maximum model usage for the price but suffer from a poor user interface, especially for those not accustomed to the command line. A hybrid solution, like a supercharged terminal with a better UI and integrated lightweight editor, aims to bridge the gap, offering a more user-friendly experience within a familiar command-line environment.

Many AI coding solutions are expanding beyond their initial focus, offering integrated features like GitHub apps for code reviews, CLI tools, and web-based background task execution, creating broader ecosystem plays. Ultimately, the best choice depends on individual preferences, whether one prioritizes graphical interfaces, command-line proficiency, or maximum usage of a specific AI model.

Short Highlights

AI code editors differ primarily in user experience, interface, and features for the price, rather than underlying LLM performance.
Dedicated AI IDEs offer superior UI integration and features like granular change acceptance but may provide less model usage due to intermediary costs.
Plugins for existing IDEs are generally less robust than dedicated solutions, while CLI tools offer maximum model usage directly from creators but have a poor UI.
Hybrid solutions, like supercharged terminals, aim to combine the benefits of both CLI and IDEs, offering a better UI within a terminal environment.
The speaker utilizes a hybrid approach, paying for Cursor ($20 subscription for tab-to-complete), Claude Code ($100/month plan), and Warp Turbo ($50/month plan) to leverage different models and features.

Key Details

AI Code Editors: Overview and Core Differentiators [00:00]

The speaker tested numerous AI code editors to help users find the best fit for their workflow.
Many AI code editors are more similar than different, often using the same LLMs.
Key differentiating factors include user experience, user interface, and the features offered for the price.
The choice of AI coding tool is a personal experience, and what works best for one person may not for another.

The biggest way that all these AI co-editors differentiate from one another is their user experience and their user interface as well as the different features that they offer for the price that you pay.

Dedicated AI IDEs [01:21]

Examples include Cursor, Windsurf, VS Code with GitHub Copilot integration, and Trey.
The primary advantage is the superior, integrated user interface with features like a dedicated chat window.
Users can review and accept individual code changes or apply them all at once.
These editors allow tagging different folders and files within the chat interface.
A significant benefit is the ability to choose which AI model to use through a model picker.
A major downside is that these editors are not built by the AI model creators, leading to less model usage for the price compared to direct subscriptions.
Intermediary costs mean users pay the editor provider, who then pays the model creator, reducing overall model usage.

So, definitely from a user experience and a user interfacing perspective, all these AI code editors win without a doubt.

Plugins for IDEs [04:17]

These are extensions installed into existing IDEs like JetBrains or VS Code.
Augment Code is a prominent player, though it appears to be migrating towards a CLI implementation.
These integration and plugin products can be considered a "watered down" version of dedicated AI IDEs.
They have limitations in UI and how they present AI agent coding workflows due to not being full-fledged IDEs.
The company behind Windsurf, Kodium, initially started as a plugin before developing its own IDE.
Integration and extension products have seen a decrease in popularity, with a shift towards CLI interfaces.

And in my opinion, some of these like integration and plug-in products, they are kind of just like a watered down version of the dedicated AI idees that you that I talked about earlier in this video.

CLI Tools [06:06]

Command-line interface tools that work directly within the command line.
Claude Code is a popular example that has been part of the speaker's workflow.
These tools allow interaction with code and agentic coding changes on the fly.
Many CLI tools are built by companies that create their own AI models (e.g., Claude Code by Claude, OpenAI Codex by OpenAI, Gemini CLI by Google).
The biggest benefit is maximum model usage because users are paying the model providers directly.
The major downside is the "horrible" user interface, especially for those not comfortable living in the terminal.
Features like accepting individual changes are not easily managed within the CLI; changes often need to be reviewed in a full IDE.

The biggest benefit of using one of these CLI tools is the fact that because you are paying these model providers directly, you are essentially paying for the utmost maximum usage for that particular model.

Hybrid Solution: Terminal with Enhanced GUI [08:26]

This category bridges the gap between pure CLI tools and dedicated AI code editors.
Warp is presented as an example of this hybrid solution, acting as a supercharged terminal.
It offers the familiarity of a CLI tool but with a significantly better user interface and experience.
Warp is model-agnostic, allowing users to choose their preferred models.
It intelligently detects whether a user is entering a terminal command or initiating an agentic coding workflow.
Features include a lightweight file tree and a lightweight code editor within the terminal itself, enabling changes and reviews without opening a separate IDE.
It's described as a "terminal and IDE had a baby," designed for agentic coding workflows.

So, it has a kind of a familiarity of a CLI tool, but it also has a much better user interface, user experience for the user.

Evolving AI Coding Solutions and Ecosystem Plays [11:05]

Many tools, initially focused on one area (e.g., dedicated IDE), are expanding to offer more services.
Examples include Cursor's GitHub app for code reviews, CLI tool, and background tasks.
OpenAI Codex and Claude also offer GitHub app integrations and background agent tools.
These providers are moving towards larger ecosystem plays, allowing code changes from anywhere via various surfaces.
These additional features and bells and whistles come at a cost, meaning payments are not solely for pure model usage.

Nowadays, in 2025, a lot of these tools, even though they started off as one thing, they've kind of evolved into something much bigger and have a lot more service offerings.

Choosing the Right AI Code Editor Solution [12:37]

For GUI lovers who dislike terminals: Dedicated AI IDEs are recommended.
For terminal enthusiasts: Warp is suggested for its balance of terminal living with IDE-like features.
For users focused on a single model: Paying directly for subscriptions from model creators (e.g., OpenAI, Claude) is best for maximum usage, though CLI UI is a drawback.
Hybrid approach: Using a solution like Warp offers a good UI while still allowing model choice and the benefits of an advanced terminal.
The speaker's current setup: $20 Cursor subscription for tab-to-complete, $100/month Claude Code plan, and $50/month Warp Turbo plan.
This hybrid approach allows switching models when needed and leverages familiar, efficient tools.
User experience is identified as the biggest differentiator among AI coding solutions.