AI Coding Tools: Evolution or Revolution? - #5

Subscribe to AI Coding Weekly

Issue #5 • 2025-05-13

AI Coding Weekly

Hey readers! 👋 This week we’re diving into OpenAI’s massive $3B acquisition of Windsurf and the increasingly crowded battlefield of AI coding tools where “everyone and their mother is building coding agents.” Whether you’re team Cursor, Claude Code, or exploring open-source alternatives like Void, the AI coding assistant space is evolving faster than we can update our .gitignore files!

🧪 BETA NOTE: This newsletter is still in beta, and we’re working out the kinks. We’d love your feedback on what’s working and what could be improved! Just hit reply to share your thoughts.

This Week’s Highlights 🔥

There’s Beauty in AI — I really loved this piece by Thorsten Ball about how AI coding tools made him faster, similar to his experience with Vim - top article that captures the transformation from seeing AI as an unpredictable slot machine to appreciating it as a powerful assistant.
Thorsten Ball

OpenAI reaches agreement to buy Windsurf for $3B — Big news that raises important questions about whether Windsurf will be locked into OpenAI’s ecosystem or remain an independent tool as competition intensifies with Microsoft and Google.
swyx

GitHub Copilot: The agent awakens — GitHub introduces ‘agent mode’ and makes ‘Copilot Edits’ generally available in VS Code, allowing Copilot to autonomously iterate on code, fix errors, and suggest terminal commands.
GitHub

Notes on rolling out Cursor and Claude Code — A fascinating discussion about how developers often forget to use AI coding tools even when they’re available, raising questions about integration into workflows and the quality of AI-generated code.
Hacker News

Wave 8: UX Features + Plugins Update — Windsurf’s latest update introduces ‘Memories’ for storing important information (which seems great!), a redesigned model selector, and a ‘Continue’ button to streamline user feedback.
Windsurf Team

Transforming Engineers: How we grew AI coding adoption at Plaid — Practical insights on implementing AI coding tools at scale, with Plaid achieving 75% adoption through streamlined pilots, dedicated ownership, and positioning AI as complementary to existing workflows.
Plaid

AI Coding Tools & Updates 🛠️

Void: Open-source Cursor alternative — A nascent open-source alternative to Cursor that allows direct connection to various AI models without routing through a backend, though documentation needs improvement.
Hacker News

Trae - Collaborate with Intelligence — Is Trae catching up with Cursor and Windsurf? Their new unified Chat and Builder interface and @Agent feature for multi-agent collaboration suggest they’re making significant strides.
Trae

Zed Agentic Editing — The super-fast Rust-based IDE now integrates AI capabilities for generating and analyzing code, though I still prefer Cursor for my daily work.
Product Hunt

GitHub - t128n/cursor-linux — Unofficial installation scripts for the Cursor code editor on Linux systems, including automatic version checking and desktop integration.
trbndev

Release v0.83.1 · Aider-AI/aider — A version bump for the Aider AI project, indicating ongoing development and improvements.
Aider-AI

JetBrains IDEs Go AI — Great news for IntelliJ IDEA users as JetBrains integrates AI features for intelligent code suggestions, error detection, and automated refactoring.
amit9gupta

Research & Benchmarks 📊

AI SQL Benchmark — Useful benchmark for devs working with SQL shows Claude-3.7-Sonnet leading with a 76.82 score, followed closely by OpenAI’s O3-Mini at 76.08 in generating SQL queries from natural language.
sh_tomer

Fine-Tuning Code Completion Models: The Easy Way — An interesting approach to improving code completion models by 25-45% through fine-tuning, though I’d probably try this only if I had the budget for it.
Refact.ai

AI Test Case Generation: 2024 Guide — AI can generate thousands of test cases in hours, with Facebook’s Sapienz tool demonstrating an 80% reduction in app crashes through AI-generated tests.
MarsX

Sizing up the AI code generators — Comparative analysis of leading AI code generators shows Claude 3.7 excels at iterative feature work while Gemini 2.5 shines in UI tasks, suggesting a multi-model workflow for optimal results.
InfoWorld

Perspectives & Experiences 🧠

I built an AI code review agent in a few hours — A pretty cool practical example showing both the potential and limitations of AI code review agents, highlighting that the core challenge is making LLMs truly understand your code.
msukkarieh

The reason why i don’t use AI or even code completion — A refreshing counterpoint arguing that AI often produces ineffective code snippets and diminishes coding proficiency, similar to how phone books eroded our ability to remember numbers.
bitbasher

Knowing when to use AI coding assistants — High-level information you can share with your CTO about when AI tools excel (boilerplate, simple functions) and when human oversight is crucial (complex projects, novel scenarios).
InfoWorld

New Tools, New Flow: The Cognitive Shift of AI-Powered Coding — Explores the challenges of adopting AI coding tools and finding a new “flow state” that effectively integrates them into daily coding practices.
Scott Logic

AI-Generated Code Needs Refactoring, Say 76% of Developers — Survey reveals that while 70% of developers use AI tools weekly, 76% need to refactor at least half of the AI-generated code due to poor readability and maintainability issues.
The New Stack

Events & Industry News 📅

@aiDotEngineer announces AI Engineer World’s Fair 2025 — Top news! A dedicated conference for AI-assisted coding featuring prominent speakers like @saranormous and @simonw, happening June 3-5, 2025 in San Francisco.
@aiDotEngineer

Building Windsurf with Varun Mohan — Podcast interview with Windsurf’s CEO discussing technical challenges in creating an AI-native IDE and the importance of embracing failure and learning in innovation.
The Pragmatic Engineer

Latent Space: The AI Engineer Podcast — Check out this podcast covering the AI coding landscape, featuring discussions on tools like Claude Code, which Boris describes as “not a product as much as it’s a Unix utility.”
radio.net

Comparisons & Reviews 🔍

Comparing Modern AI Coding Assistants — A good deep dive comparing Web-Based AI Development Tools, IDE Extensions, AI-First IDEs, and CLI Agents, highlighting their unique strengths and ideal use cases.
Roberto Infante

Firebase Studio vs DeepSite vs Canva Code — Canva Code emerged as the clear winner in this comparison, successfully generating functional code on the first attempt in 80% of test cases.
AI Augmented Living

Ask HN: How much better are AI IDEs vs. copy pasting into chat apps? — Discussion highlighting the convenience of AI IDEs in understanding project context directly versus the tediousness of providing context in chat prompts.
Hacker News

Beyond file trees: why AI coding assistants need smarter context — Explores how AI tools need to move beyond basic file navigation to utilize relevant context through solutions like the Nuanced code graph.
Nuanced

Made with ❤️ by the AI Coding Weekly team. Have questions, comments, or feedback? Just hit reply - we’d love to hear from you!

Your email address is safe with us. We never share your information with third parties.

Data Drift Press, 1388 Haight St #1199, San Francisco, CA 94117, United States

This newsletter was created using AI technology and reviewed by humans at Data Drift Press.