AI-Assisted Development

Tools for AI-Assisted Coding

Workshop — Part 3: The landscape

3 / 4 — Tools
AI-Assisted Development

Four Categories

IDE plugins
Add AI into your existing editor
GitHub Copilot, JetBrains AI, Continue

AI-first IDEs
Editor built around AI from the ground up
Cursor, Windsurf

Agent / chat tools
Multi-step tasks outside the editor
Claude Code, ChatGPT, Codex

Browser builders
Spin up full apps in the browser
v0, Lovable, Bolt

3 / 4 — Tools
AI-Assisted Development

Pick 1–2 tools that fit your stack.

Learn them deeply.

You do not need all of them.

3 / 4 — Tools
AI-Assisted Development

GitHub Copilot

Type: IDE plugin
VS Code, JetBrains, Neovim

Strengths

  • Fast multi-line inline completions
  • Deep GitHub integration (PR summaries, diffs, docs)
  • Minimal friction — install, log in, code

Best when:

  • Speeding up daily coding without changing editor
  • Team is GitHub-centric

Keep in mind:
Less suited for large multi-file refactors (but improving)

3 / 4 — Tools
AI-Assisted Development

Cursor

Type: AI-first IDE (VS Code fork)

Strengths

  • Project-wide edits and large refactors
  • Agent mode: multi-step tasks, fixes lints and tests
  • Multi-model: OpenAI, Anthropic, Google
  • .cursorrules for team-wide standards

Best when:

  • Full-stack work touching many files
  • Standardizing AI usage across a team

Keep in mind:
Invest time in project configuration for best results

3 / 4 — Tools
AI-Assisted Development

Claude / Claude Code

Type: Autonomous coding agent

Strengths

  • Long-context reasoning across large projects
  • Architecture reviews, migrations, hard bugs
  • CLI, VS Code, JetBrains, Slack, CI/CD integrations

Best when:

  • “AI pair architect” for design decisions
  • Analyzing or modernizing large codebases

Keep in mind:
Best paired with an IDE, not as your only tool

3 / 4 — Tools
AI-Assisted Development

Windsurf & ChatGPT

Windsurf — AI-first IDE (Codeium)

  • Aggressive multi-line Tab-to-Jump completions
  • Cascade agent — searches docs, automates tasks
  • VS Code, JetBrains, Vim, Jupyter

Best for: fast completions + agent workflows

ChatGPT — General-purpose assistant

  • Great for learning, walkthroughs, explanations
  • Code Interpreter — run Python in chat
  • Custom GPTs for domain-specific helpers

Best for: exploring new tech, prototyping outside your repo

3 / 4 — Tools
AI-Assisted Development

Browser Builders

v0 (Vercel)
Design-to-code for React / Next.js
Instant browser preview → one-click export
Best for: beautiful frontends fast

Bolt
Browser-based full-stack generation
Best for: quick proof-of-concepts

Lovable
Chat-to-app full-stack builder
Describe your vision → working app
Best for: MVPs and internal tools

All three are great for prototyping — export to a real repo when you need production-grade control.

3 / 4 — Tools
AI-Assisted Development

Open & Flexible Options

Continue — Open-source AI extension

  • Bring your own model (OpenAI, Anthropic, Llama, on-prem)
  • VS Code and JetBrains
  • Best for strict security / compliance requirements

Codex — Open-source CLI agent (OpenAI)

  • Similar to Claude Code but fully open-source
  • Self-hostable for privacy or compliance

JetBrains AI — Native JetBrains AI layer

  • Deep IDE awareness, JVM / enterprise stacks
  • Best if your team is standardized on JetBrains
3 / 4 — Tools
AI-Assisted Development

Quick Comparison

Tool Type Excels At
GitHub Copilot IDE plugin Daily coding, VS Code / JetBrains
Cursor AI-first IDE Full-stack work, large refactors
Claude Code Autonomous agent Big projects, analysis, architecture
Windsurf AI editor Fast completions, agent workflows
ChatGPT Chat assistant Learning, design, exploration
v0 / Lovable Browser builder Prototypes, MVPs, design-to-code
Continue Open-source plugin BYO models, privacy, flexibility
JetBrains AI JetBrains-native JVM / enterprise teams
3 / 4 — Tools
AI-Assisted Development

How to Choose

Stay in my current editor?
→ IDE plugin: Copilot, Continue, JetBrains AI

Full-stack or multi-file work?
→ AI-first IDE: Cursor, Windsurf

Large codebase analysis / architecture?
→ Agent: Claude Code, Codex

Fast prototype or UI?
→ Browser builder: v0, Lovable

3 / 4 — Tools
AI-Assisted Development

No tool replaces human judgment

on design, security, and correctness.

3 / 4 — Tools
AI-Assisted Development

Summary

  • Many tools — but most share the same core AI capabilities
  • Pick 1–2 and invest in learning them properly
  • Use context files (.cursorrules, AGENTS.md) for consistent results
  • Match the tool to the task — don’t over-engineer your toolchain

Next: The problems and failure modes to watch out for →

Navigate with arrow keys · F for fullscreen