Agentic Workflows: Technical Skills & Toolset

What Is This Program?

Students learn to build "Autonomous Systems"—multi-agent frameworks where specialized AI entities collaborate, use external tools, and self-correct to achieve complex, long-term goals.

Who Is It For?

Developers and system architects aged 16+ interested in AI autonomy and orchestration.

What Will I Learn?

Role-Based collaboration (CrewAI), Stateful agents (LangGraph), Model Context Protocol (MCP), and Agent-Friendly search (Tavily).

Pre-Requisites

Familiarity with process logic (Decomposition, Conditional Logic), Key-Value pairs, Few-Shot & Chain-of-Thought.

Dates

Session 1: July 6 - July 17, 2026
Session 2: August 3 - August 14, 2026 (EN)

Language

This course is offered in both English or Chinese. Check the dates for your language preference.


Tuition & Fees

Tuition: NT$XX,000
Deposit: NT$2,000
Early Bird Deal: Save 15% (Book by March 1st)

Core Tech Stack

CrewAI

The primary framework for "Role-Based" multi-agent collaboration (best for team-like workflows).

LangGraph

A graph-based framework for building "Stateful" agents that need complex branching and looping logic.

Model Context Protocol (MCP)

The universal standard for connecting agents to local data, tools, and remote APIs safely.

Tavily / Search APIs

Specialized "Agent-Friendly" search engines that return clean, LLM-ready data rather than raw HTML.

Curriculum

Week 1: Agent Architecture & Role Design

Day Topic Hands-On Activity
01 Environment & MCP The Agent Gateway: Setting up Python/Poetry and using MCP.
02 Persona Engineering Role Definitions: Learning to define "Backstories" and "Goals".
03 Task Decomposition The Breakdown: Breaking a massive goal into executable steps.
04 Orchestration Patterns Flow Design: Sequential vs. Hierarchical workflows.
05 Multi-Agent Collaboration The Crew Launch: Building a three-agent "Crew".

Week 2: Tool Use, Memory & State

Day Topic Hands-On Activity
06 Function Calling The Tool Belt: Defining typed interfaces for LLM calls.
07 Error Handling & Fallbacks Self-Correction: Designing logic for when tools fail.
08 Human-in-the-Loop (HITL) The Breakpoint: Implementing pauses for user approval.
09 Short & Long-Term Memory The Vector Archivist: Integrating Pinecone or Chroma.
10 State Persistence The Pause/Resume Lab: Using LangGraph to save state.
Enroll in Agentic Track