Product Builder: Agentic Skills
Ryan Smith
Skillsmith Founder/Shopify/Delphia YC18
Build agentic AI systems that survive real project complexity
Agent skills are where quality, speed, and cost of your projects make or break. Learning agent skills will level up your product building.
What Students Leave With
A production-quality SKILL.md
A multi-agent system that delegates to 2+ specialists
The Agent-Skill Matrix framework
The Token Budget Calculator
The Delegation Architecture pattern
The Composition Checklist
The Anti-Pattern Detection Checklist
What you’ll learn
Learn agent-skill composition to go from building AI demos to shipping AI products teams depend on.
Apply the agent-skill matrix to classify every component in your workflow
Extract procedural knowledge from agent files into modular, reusable skills
Use the composition checklist to audit existing setups for mixed concerns
Calculate token budgets across agents, skills, and conversation history
Implement scoped delegation so each agent operates under 20K tokens
Recognize when output degradation signals context overflow and fix it
Design orchestrator-specialist-reporter workflows with clear handoff points
Implement the daisy-chain pattern where skills trigger their own sub-agents
Apply dependency inversion so agents depend on abstractions, not implementations
Learn directly from Ryan
Ryan Smith
Founder of Skillsmith, x-Shopify, Delphia (YC18), and deep learning founder
Who this course is for
Technical leads adopting AI-assisted development who need a decision framework before scaling agent
tooling across their team.
Engineers building multi-agent workflows who need practical patterns for context management, delegation, and composability.
Product managers/designers using AI coding tools who want to move beyond prompting into scalable agent architecture for real products.
Prerequisites
What experience is needed for this course?
Three or more months using Claude Code, Cursor, Antigravity, Codex or similar agentic framework.
Tokens are not included in the course fee
We will likely exceed $20/month plans with Anthropic, OpenAI, Google, etc. Come prepared with your own API key or $100/mo subscription
Course syllabus
2 live sessions • 12 lessons
Week 1
Schedule
Live sessions
4 hrs / week
Two 4-hour sessions with a bonus hour for cohort learning
Thu, Mar 5
6:00 PM—10:00 PM (UTC)
Thu, Mar 12
5:00 PM—9:00 PM (UTC)
Projects
2-4 hrs / week
The learning comes from building. You will want to build at least one skill per week after the live session.
$899
USD