Major Update Β· 8 Lightweight Apps Β· Zero Code Onboarding

Workshop

Never written a line of code? No problem. We've redesigned the AI Agent workshop so anyone can start β€” from using ready-made AI in 5 minutes, building your own Bot in 30 minutes, to running a complete project in 1 hour.

🚦 Pick Your Starting Point

Which type of learner are you?

Choose based on your background. Each path ultimately leads to independently building a large-scale Agent project.

Lab 01 Β· Zero-Code Onboarding

No code, 5 minutes to experience AI Agents

No programming background required. Three progressive stages: use ready-made, build your own, then run complete code for the first time.

01.1 Β· 5 Minutes

🌟 Use ready-made AI Agents to get things done

No setup needed. Just open a browser and start talking to AI. The key is to use it as your assistant to solve real problems.

πŸ“‹ 6 Real Scenarios (try 1-3)

πŸ“š Aid reading long articles

Paste a 5000-word industry article, ask for: 3 core points, 5 hard terms, next actions.

πŸ’‘ Explain concepts

Use 3 levels of analogy to explain Transformer attention to a 10-year-old.

🎯 Build a study plan

Ask for a 4-week LangGraph plan with 1.5 hours daily, with tasks and checklists.

πŸ” Decision aid

Have AI play senior HR and grill you 5 questions you might have missed.

πŸ“ Writing collaboration

Paste your draft, ask for 3 unclear spots, 3 examples to add, and a better opening.

πŸ§ͺ Brainstorming partner

Got a vague idea? Ask for 10 common solutions and the trade-offs.

01.2 Β· 30 Minutes

πŸ›  Build Your First Dedicated AI Agent

πŸ“Š Mainstream Platforms Compared

PlatformDifficultyFree TierBest For
ChatGPT Custom GPTs⭐ EasiestPlus subscriptionPersonal knowledge Q&A
Claude Projects⭐ EasiestPro subscriptionLong document research
CozeEasyFree tier sufficientBots published to IM
DifyEasyCloud SaaS freeKnowledge base + workflows
FlowiseMediumFully OSS & freeDeep customization
n8n + AI nodesDeeperSelf-hosted freeComplex business automation

πŸš€ Recommended Path A: 5 minutes with ChatGPT Custom GPT

Step 1: Open ChatGPT

Visit chatgpt.com and log in. Requires ChatGPT Plus ($20/mo).

Step 2: "Explore GPTs" β†’ "Create"

Click "Explore GPTs" on the left, then "Create" in the top right.

Step 3: Configure with natural language

Type: "You are a learning assistant. I'll give you PDF notes; please answer questions based on them with original citations."

Step 4: Upload materials to "Knowledge"

Upload PDFs or articles you want to query.

Step 5: Save & use

Click save. Select this GPT in the chat and ask any question about your material.

01.3 Β· 60 Minutes

🐣 First Complete Code Run (no need to understand)

πŸ“‹ Environment Setup (10 min)

Option 1: Local install (recommended)
  1. Install Python 3.10+: python.org/downloads
  2. Install VS Code: code.visualstudio.com
  3. Create folder my-first-agent
  4. Open it with VS Code
Option 2: Online (fastest)
  1. Open Replit.com
  2. Create new Python Repl
  3. Paste the code and run
  4. No local install needed

πŸ“‹ Get an API Key

OpenAI

platform.openai.com
New users get $5 free (~200+ calls)

DeepSeek

platform.deepseek.com
Fast in China, very cheap

Zhipu GLM

open.bigmodel.cn
Free tier, strong Chinese

agent.py Β· Copy and run
# Your first AI Agent β€” copy and run, no need to understand
#
# How to use (4 steps):
#   1. Install deps:  pip install openai
#   2. Set key:       export OPENAI_API_KEY="your-key"
#   3. Run:           python agent.py
#   4. Ask questions: type anything, press Enter

import os
from openai import OpenAI

client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

SYSTEM_PROMPT = """You are a patient AI assistant.
- Ask for clarification when questions are unclear
- Keep answers concise
- Respond in English""".strip()

print("πŸŽ‰ Your first AI Agent is running!\nType 'quit' to exit\n")

while True:
    question = input("You: ")
    if question.lower() in ["quit", "exit"]:
        print("πŸ‘‹ Bye!")
        break

    response = client.chat.completions.create(
        model="gpt-4o-mini",  # cheap & capable
        messages=[
            {"role": "system", "content": SYSTEM_PROMPT},
            {"role": "user", "content": question},
        ],
    )
    print(f"AI: {response.choices[0].message.content}\n")
Run command (copy to terminal)
# Step 1: Install deps
pip install openai

# Step 2: Set key (Mac/Linux)
export OPENAI_API_KEY="sk-..."

# Windows PowerShell:
$env:OPENAI_API_KEY="sk-..."

# Step 3: Run Agent
python agent.py
01.4 Β· CORE Β· Lightweight App Projects

πŸš€ 8 AI Agents You Can Use Right Now

This is the core of Agent Academy. Each project actually solves a real daily problem. Each provides 3 implementation paths.

Path A Β· No Code

Build via ChatGPT Custom GPT / Coze / Dify GUI. For those who don't want to touch code at all.

Path B Β· Visual Platform

Drag-drop with Dify / Coze / Flowise. For more complex Agents without code.

Path C Β· Write Code

50-100 lines of complete runnable Python. For deep understanding.

πŸ“¦ 8 Projects at a Glance

πŸ“°

Project 1 Β· Daily Learning Digest

Aggregate RSS/news content, generate daily digest

πŸ“š

Project 2 Β· Personal Knowledge Q&A

Upload PDFs/notes, query with natural language

⏰

Project 3 Β· Time-Block Planner

Input available time + tasks, get optimal schedule

πŸ’Œ

Project 4 Β· Email/IM Reply Assistant

Forward a message β†’ AI generates reply drafts

πŸ“

Project 5 Β· Weekly Report Generator

Paste bullet points β†’ structured report

πŸ”

Project 6 Β· Information Verification

Multi-angle analysis of article credibility

🎴

Project 7 Β· Study Card Generator

Convert notes to Anki-style flashcards

πŸ’Ό

Project 8 Β· Job Interview Coach

Input job description β†’ mock interviews + resume tips

βœ“
Three-Path Trade-off

Path A: fastest but personal use only Β· Path B: publishable Β· Path C: most control

Lab 02 Β· Framework Mastery

Use mainstream frameworks for real apps

Lab 2.1 Β· LangChain Knowledge Q&A

Turn a product manual into a Q&A assistant. Learn the full RAG chain: load β†’ split β†’ embed β†’ retrieve β†’ generate.

RAGLangChain

Lab 2.2 Β· LangGraph Research Assistant

Input a topic β†’ search β†’ analyze β†’ output report. Master state machines, interrupt, human-in-loop.

LangGraphState Machine

Lab 2.3 Β· CrewAI Multi-role Writing

Input topic β†’ 3 Agents collaborate (research, write, review) β†’ article. Learn role-based collaboration.

CrewAIMulti-Agent

Lab 2.4 Β· AutoGen Dialogue Coding

User + engineer + reviewer dialogue solves coding problems. Microsoft-style dialogue-driven dev.

AutoGenDialogue
Lab 03 Β· System Design

From toy to production

Lab 3.1 Β· Long-term Memory Chatbot

A bot that remembers your preferences, what you've said, past questions.

Lab 3.2 Β· Plan-and-Execute Planner

Input "research AI Agent industry" β†’ auto-plan 5 steps β†’ execute β†’ output report.

Lab 3.3 Β· Observability Integration

Integrate LangFuse / LangSmith to see every tool call and token consumption.

Lab 04 Β· Evaluation & Optimization

Make the Agent run well and cheap

Lab 4.1 Β· RAGAS evaluation for your knowledge base

Lab 4.2 Β· DSPy auto prompt optimization

Lab 4.3 Β· Tool ecosystem (Composio)

Lab 05 Β· Complete Project

From 0 to a production-grade Agent

πŸ›’ 5.1 Β· Smart Shopping Assistant

Multi-Agent collaboration: search + price compare + review + final recommendation.

πŸ“š 5.2 Β· Company Document Q&A

RAG + permission management, multi-source index, citation tracking.

πŸ”¬ 5.3 Β· Automated Research Assistant

Plan-and-Execute: literature retrieval + summary + draft.

πŸ’Ό 5.4 Β· Sales Automation

Lead management, email analysis, CRM sync.

πŸ’» 5.5 Β· Coding Agent

GitHub integration: issues, code, tests, PRs.

πŸŽ“ 5.6 Β· Education Tutor

Personalized: diagnose gaps, custom problems, progress tracking.

Done with labs? Next:

Ready to deliver a complete large-scale Agent project. Blueprint has detailed design templates and code skeletons.

View Agent Blueprint β†’