Skip to main content

Zed's Agentic Editing: AI-Native Code Editing with Nathan Sobo

ChangelogMay 7, 20251h 41min3,753 views
33 connections·40 entities in this video→

Zed's Agentic Editing Experience

  • πŸš€ Zed is evolving into an AI-native code editor, enabling collaboration with agents at 120fps within a multiplayer IDE.
  • πŸ’‘ The new agentic experience focuses on a conversational interface optimized for understanding and editing code, moving beyond a low-level LLM integration.
  • 🧩 Tool calling is integrated, allowing agents to access the codebase and perform streaming edits, with changes reviewable in a multibuffer.
  • πŸ› οΈ Zed's multibuffer acts as a single, coherent virtual buffer composed of content from multiple files, enhancing the editing experience.
  • πŸ’¬ The agentic interaction builds a curated, editable subset of the codebase, presenting a new paradigm for code collaboration.

Navigating AI in Software Development

  • 🧠 Zed is built from first principles, allowing for deep integration of new technologies like AI, despite a potentially slower initial rollout.
  • ⚠️ Testing inherently stochastic AI systems presents a unique challenge, requiring new approaches to ensure reliability and predictability.
  • πŸ’‘ The functional core, imperative shell paradigm is discussed, with AI potentially acting as the 'shell' around Zed's deterministic core.
  • 🎯 Prompt engineering is crucial, with examples showing how subtle prompt adjustments can significantly improve AI output reliability.
  • πŸ“ˆ The goal is to elevate the quality of software by leveraging AI to automate tedious tasks, allowing developers to focus on higher-level problem-solving.

The Future of Code Editing and Business Models

  • ⚑ AI is seen as a catalyst for accelerated software development, enabling faster iteration and experimentation.
  • πŸš€ Zed aims to be the first AI-native editor, built on a solid foundation of core technology, offering a differentiated experience.
  • πŸ’° While Zed remains open-source, new AI services will be offered as a paid subscription, providing convenience and compute resources.
  • 🀝 The business model emphasizes developer control and choice, allowing users to bring their own API keys or leverage Zed's hosted services.
  • 🌐 Zed's vision extends to a vertically integrated collaborative environment where human and AI interactions seamlessly occur within the editor.
Knowledge graph40 entities Β· 33 connections

How they connect

An interactive map of every person, idea, and reference from this conversation. Hover to trace connections, click to explore.

Hover Β· drag to explore
40 entities
Chapters19 moments

Key Moments

Transcript373 segments

Full Transcript

Topics15 themes

What’s Discussed

Agentic EditingAI-Native EditorZed IDECode CollaborationLarge Language Models (LLMs)Prompt EngineeringFirst PrinciplesSoftware DevelopmentDeveloper ExperienceMultibufferTool CallingOpen SourceBusiness ModelsAI ServicesCompute
Smart Objects40 Β· 33 links
ProductsΒ· 19
PeopleΒ· 4
ConceptsΒ· 13
CompaniesΒ· 2
MediasΒ· 2