Can AI Really Boost Your Front‑End Coding? A Hands‑On Comparison of VS Code IntelliCode, Kite, TabNine, and aiXcoder

There is no doubt that artificial intelligence will change the way software is developed – we have already seen benefits such as automatic code generation from images or data models. This article investigates whether an ordinary developer can already use useful AI tools to improve productivity.

Experiment Setup

The author selected common programming tools labeled as "intelligent programming" and attempted a simple experiment: creating an interface with an input box whose content updates as the user types, to observe the capabilities each tool provides.

VS Code IntelliSense Overview

VS Code is the most popular editor among front‑end developers and offers built‑in intelligent features that can be experienced quickly.

IntelliSense Components

Code Completion : suggestions based on language semantics and source analysis, displayed as a list that filters as you type.

Quick Info : hover over a suggestion to view documentation in a side pane.

Parameter Info : after selecting a function, shows its signature and highlights the current parameter.

The suggestion list includes language service recommendations, snippets, global identifiers, and simple word‑based completions, sorted in that order. The locality bonus (disabled by default) promotes suggestions that appear nearby in the same file; it can be enabled via the editor.suggest.localityBonus setting.

Snippet suggestions are controlled by editor.snippetSuggestions, which can be set to "none", "top", "bottom", or "inline" (default).

The default editor.suggestSelection prefers previously used items, but can be changed to first, recentlyUsed, or recentlyUsedByPrefix.

Code Actions

Beyond suggestions, VS Code provides actionable features such as refactoring (extract function/variable) and quick fixes (e.g., adding missing this, fixing misspelled properties, removing unused imports). These are triggered via the light‑bulb icon or keyboard shortcuts.

Tool Reviews

Visual Studio IntelliCode

IntelliCode, released by Microsoft in 2018, supports multiple languages and has over six million downloads. It offers smarter code completion by ranking suggestions based on usage patterns, parameter completion, and refactoring recommendations. However, in the author’s front‑end experiment, its completion efficiency was similar to native VS Code.

Kite

Kite, launched in 2017, focuses on Python but also supports JavaScript. It provides line‑level completions, multi‑line suggestions, and context‑aware recommendations. In the experiment, Kite’s completions were richer and more accurate than the previous tools.

TabNine

TabNine, originally open‑source and now owned by Codota, covers many languages. It offers full‑type completions, including single‑token, multi‑token, line, and multi‑line suggestions. While its suggestion frequency is high, accuracy is slightly lower than Kite, though it displays confidence percentages without noticeable latency.

aiXcoder

aiXcoder, a Chinese product from Peking University, supports Java, JavaScript, and Python. It provides token‑level, multi‑token, line, and multi‑line completions, as well as intelligent code search, documentation search, and project‑level quality analysis. The personal version only supports Java, so the author could not test it on the front‑end example.

Other Tools

DeepCode – detects bugs, security issues, and performance problems.

Intelli Refactor – enhances VS Code’s built‑in refactoring.

Capability and Experience Comparison

The author presents tables (images) comparing the coverage of features and user experience metrics such as completion efficiency, accuracy, response speed, and offline resource usage. A disclaimer notes that the experiment is limited to front‑end code and personal setup, so results are for reference only.

Is It Just Hype?

The conclusion is that AI‑assisted programming is still largely marketed hype, but the tools already provide measurable productivity gains, especially for junior developers. For senior developers, the benefit is smaller and may become a distraction unless highly customizable. The effectiveness depends on the quality of training data, and future developments may include programming robots that collaborate with developers.

Author: Core developer of Iceworks, a multi‑platform development suite.