AI Productivity by Role
This matrix shows how the 5 levels apply across different roles. Use this as inspiration. Roles blend, and you may find ideas from multiple columns relevant to your work.
The Matrix
| Level | Developers | QA Engineers | Designers | Business Analysts | Project Managers | Tech Leads |
|---|---|---|---|---|---|---|
| 1. Chat | Ask syntax questions, explain errors, generate snippets | Research test strategies, brainstorm test cases | Explore design ideas, get copy variations | Draft requirements, clarify terminology | Draft communications, summarize notes | Research approaches, get second opinions |
| 2. Context | Paste code for review, debug with full error output | Upload bug screenshots, analyze test logs | Share mockups for accessibility review | Attach docs for gap analysis | Share timelines for risk analysis | Upload diagrams for architecture review |
| 3. Connected | IDE integration (Cursor, Copilot), CLI tools (Claude Code) | Test generators connected to framework | Figma plugins, design system tools | Jira/Confluence AI integration | PM tool AI features, meeting transcription | MCP servers, codebase analysis tools |
| 4. Customized | CLAUDE.md with conventions, custom commands, hooks | Test templates matching your framework | Design system rules in prompts | Domain glossaries, requirement templates | Custom reporting prompts, status templates | Team-wide context files, review checklists |
| 5. Autonomous | AI implements features across files with review | AI generates and maintains test suites | AI explores design variations systematically | AI drafts requirement sets from inputs | AI handles routine reporting with oversight | AI coordinates refactoring, pre-reviews PRs |
Tips by Role
Developers
- Let AI review your code before human review
- Learn your IDE tool's features. Most go unused.
- Build trust with autonomous tasks incrementally
QA Engineers
- Start with test case generation, which is high value and low risk
- Encode domain knowledge for better test suggestions
- Use AI for test maintenance, not just creation
Designers
- Always provide user context for relevant feedback
- Use AI for exploration, your judgment for decisions
- Document your design system for AI to reference
Business Analysts
- Build domain glossaries AI can reference
- Use AI for first drafts, apply expertise to refine
- Cross-reference requirements for consistency
Project Managers
- Don't let AI draft sensitive communications without editing
- Automated reports still need your review
- Keep human connection. Don't automate relationships.
Tech Leads
- Create shared context files for team consistency
- Model effective AI usage for your team
- Use AI to scale your impact, not replace your judgment