Zed's NEW Agent Mode: RIP Cursor & Windsurf! This OPENSOURCE AI Editor JUST LEVELED UP!

Zed Editor's New Agentic Mode: A Detailed Overview

Key Concepts:

• Zed: Open-source, Rust-based code editor with built-in AI features.
• Agentic Mode: AI system that can independently determine which files to read/write based on a prompt.
• Profiles: Customizable configurations that control which tools the AI has access to (e.g., read-only, code writer).
• MCP (Model Compute Provider) Server: Allows integration of various AI models (e.g., Llama, DeepSeek, Anthropic, Gemini, Mistral).
• Ninja Chat: An all-in-one AI platform that provides access to top AI models like GPT-4o, Claude 3.7 Sonnet, and Gemini 2.0 Flash.

1. Introduction to Zed and its AI Capabilities:

• Zed is an open-source editor created by the developers of Atom, built from scratch in Rust.
• It's designed with AI in mind, offering features like inline assistance and a chat panel.
• Zed is known for its speed and low memory consumption, rendering at approximately 120 fps.
• The video focuses on Zed's new "agentic mode," which allows the AI to autonomously handle tasks based on a single prompt.

2. The Need for Agentic Capabilities:

• Traditional AI coding tools require explicit instructions for each step.
• Agentic mode enables the AI to "vibe code," meaning it can figure out the necessary actions (reading/writing files) based on a high-level prompt.
• This is similar to tools like Klein or Windsurf, but Zed's implementation aims for better performance and integration.

3. Zed's Agentic Mode: Early Access and Features:

• The presenter received early access to Zed's agentic mode and highlights its rapid development.
• The new UI in the assistant panel allows users to type prompts and attach context manually (files, URLs).

4. Profiles: Controlling AI Access and Permissions:

• Profiles are a key feature for managing the AI's capabilities.
• Two default profiles: "code writer" (full access) and "read-only" (limited access).
• Profiles control which tools the AI can use (e.g., viewing diagnostics, fetching content, listing directories, writing code).
• Users can create custom profiles in the settings file, with a graphical interface planned for future updates.

5. Model Selection and MCP Servers:

• Zed allows users to select the AI model they want to use (Anthropic, OpenAI, Llama, DeepSeek, Gemini, Mistral).
• Users receive $10 of credit initially and can top up at standard rates.
• Integration with MCP servers allows users to connect to various AI model providers, similar to Klein's marketplace.

6. Demonstration: Initializing a Next.js App and Creating an Image Cropper:

• The presenter demonstrates Zed's agentic mode by asking it to initialize a Next.js app and create a simple image cropper tool.
• Zed successfully initializes the project, creates necessary files, and structures the application.
• The presenter notes that the lack of streaming responses is a downside but expects it to be added in future updates.

7. Git Integration and Version Control:

• Zed primarily uses Git for version control.
• Users can view code changes in the diff view, stage changes, and commit them directly from the editor.
• A non-Git option is also under development.

8. Performance and Advantages of Zed's Agentic Mode:

• Zed's agentic mode is described as "snappier" than tools like Windsurf or Cursor.
• It consumes less memory than VS Code.
• The tight integration of AI from the ground up results in faster response times.
• The presenter praises Zed's transparent pricing, open-source nature, and the ability to use custom API keys or local models.

9. Ninja Chat: An All-in-One AI Platform:

• Ninja Chat is introduced as a platform providing access to multiple AI models (GPT-4o, Claude 3.7 Sonnet, Gemini 2.0 Flash) for $11/month.
• Features include an AI playground for comparing model responses and a mind map generator.
• The basic plan includes 1,000 messages, 30 images, and 5 videos monthly.
• Discount codes "king25" (25% off any plan) and "king40yearly" (40% off annual subscriptions) are provided.

10. Conclusion:

• The presenter is highly impressed with Zed's new agentic mode, praising its performance, features, and integration.
• They express a strong likelihood of switching to Zed once it's more fleshed out.
• Zed's open-source nature, transparent pricing, and support for various AI models make it a compelling option for developers seeking AI-powered coding assistance.