Stop writing monolithic prompts and hoping for the best. This full-day training teaches systematic layering for AI-assisted development: decomposing complex interactions into skills (reusable capabilities), agents (specialized roles), tasks (sequential workflows), and processes (orchestrated systems). Drawing from Elixir’s compositional patterns, you’ll learn to engineer predictable AI workflows instead of crossing your fingers.
You’ll discover how to get rapid feedback at each layer rather than waiting until the end, and how to anchor long-running workflows in markdown documents and code comments that persist context across sessions. Through hands-on exercises building Phoenix applications, you’ll transform sprawling prompts into debuggable architectures where each piece can be understood, tested, and iterated independently.
Leave with practical patterns for reliable AI-assisted development that feel as solid as the OTP patterns you already trust.
What You’ll Build:
- Layered prompt architecture for Phoenix applications
- Specialized agents for development phases
- Reusable skills library
- Workflows with feedback checkpoints
- Markdown and code-based workflow anchors
What You’ll Learn:
- Decomposing prompts into testable layers with fast feedback
- When to create skills vs. agents vs. tasks
- Using artifacts to anchor workflow state
- Debugging by isolating layers
Target Audience: Elixir developers seeking systematic AI-assisted development approaches. Applicable whether using Claude Code, Cursor, or building custom AI integrations.
Takeaway: A framework for treating AI interactions as architecture—composed of understandable, controllable layers you can reason about independently. Move from unpredictable experimentation to engineered workflows with early feedback and persistent anchors.