Level 3: Connected
"AI lives inside my development environment."
What This Level Means
At Level 3, AI is connected to your work environment. It can read your files, understand your project structure, and work with your actual codebase. No more copy-paste. AI is embedded in your workflow.
This is the transition from "using AI" to "working with AI."
Characteristics of Level 3:
- AI tools integrated into IDE/editor
- Direct access to codebase and project files
- Can read and write files with permission
- Understands project context automatically
- Real-time assistance as you work
The Leap from Level 2
| Level 2: Context | Level 3: Connected |
|---|---|
| Paste code into AI | AI reads your codebase |
| One file at a time | Full project understanding |
| Context is what you provide | Context is automatic |
| Conversations are separate | Assistance is inline |
| Copy suggestions back | AI applies changes directly |
Common use cases:
- Asking questions that require reading multiple files
- Getting AI assistance inline as you work
- Having AI apply changes directly instead of copy-paste
- Analyzing patterns across your entire project
Getting to This Level
IDE Integration (Developers)
- Cursor: VS Code fork with built-in AI
- GitHub Copilot: Works in VS Code, JetBrains, etc.
- Cody: Sourcegraph's AI with codebase understanding
- Windsurf: AI-first IDE by Codeium, focused on agentic workflows and large-scale codebase refactoring
CLI Tools (Developers)
- Claude Code: Anthropic's agentic CLI tool
- Use in terminal alongside your editor
- Full read/write access to your project
MCP (Model Context Protocol)
- Connect AI to external tools and data sources
- Database access, API documentation, internal tools
- See MCP documentation
Tool-Specific AI Features
Look for AI integrations in your primary tools:
- Design: Figma AI plugins, Adobe Firefly integration
- PM: AI features in Jira, Asana, Monday.com
- Documentation: Notion AI, Confluence AI
- Communication: Teams/Slack AI integrations
Common Pitfalls
"I gave AI access but don't use it differently."
Connected AI changes what's possible. Ask questions you couldn't before: "Find all places we handle authentication" or "What's our error handling pattern?"
"I'm worried about AI accessing sensitive code."
Review your tool's privacy policy. Many enterprise tools don't train on your code. For highly sensitive projects, consider air-gapped solutions.
"The suggestions don't match our conventions."
Connected AI is powerful but generic. Level 4 (Customized) addresses this by teaching AI your specific patterns.
"I still copy-paste from chat."
Build new habits. When you catch yourself opening browser AI, stop and use your connected tool instead.
Ready for Level 4?
You're ready to progress when:
- AI is integrated into your primary work environment
- You ask questions that require reading multiple files
- You want AI to know your project's conventions automatically
- You're repeating the same context/instructions across sessions
Continue to Level 4: Customized →
The Leap Ahead
| Level 3: Connected | Level 4: Customized |
|---|---|
| AI sees your code | AI knows your conventions |
| Generic assistance | Project-specific guidance |
| Same AI for everyone | AI tailored to your project |
| Repeat instructions | Instructions encoded once |