Agent Building for PMs Pt. II

​Part II: Agent Building Deep Dive

​You built the agent. Now let's make it actually work.

​Part I gave you the foundation — Claude Code, CLAUDE.md, MCP tools, a working agent. Part II is about making it fast, reliable, and something your team can use without you.

​What we'll cover:

• ​Sub-agents — spawn parallel agents for parallel tasks, keep context clean per agent

• ​Hooks — rules that run before and after every action (block bad behavior, log what it touched)

• ​Skill files — give your agent a style guide it follows without being asked

• ​Memory — make it remember context across sessions, not just within one

• ​Scheduled runs — trigger your agent on a timer, not just when you type

• ​Packaging — turn what you built into something a teammate can clone and run in one command

​You'll walk out with an agent that runs on its own, follows your rules, and gets smarter over time.

​Prerequisites:

• ​Completed Part I (or equivalent setup)

• ​Claude Code installed and working

​Details:

• ​Online via Zoom

• ​This will be recorded

• ​Hosted by Alon Binman, Founder and CEO of Deway.ai

• ​A collaboration between Deway.ai x Product Advisory Studio

​No theory. Just the parts that change how you work.