Hands-on lab workflows and web apps for building agentic AI automations with n8n — from form-to-email flows to a Retrieval-Augmented Generation (RAG) chatbot grounded in your own documents.
📘 Course Page · 📖 Step-by-Step Guide · 🐛 Report Bug · 💡 Request Feature
Note
These are the official hands-on lab materials for the WSQ course:
Course Code: TGS-2023035977 · by Tertiary Courses / Tertiary Infotech
Course page: https://www.tertiarycourses.com.sg/wsq-agentic-ai-automation-with-n8n.html
Activity 1 — Flyer with QR Code · Form Trigger → Gmail + a QR code on an event flyer.
Activity 2 — Capture Submissions in a Data Table · Every form submission saved to an n8n Data Table alongside the email.
Activity 3 — Conditional Response · IF-node routing: "Yes" saves to a Data Table (3a) / Google Sheets (3b); "No" sends a thank-you email.
Activity 4 — Telegram AI Agent · Telegram-triggered AI agent with memory (4a) + a Data Table tool for HR lookups (4b).
Activity 5 — Website Chatbot via Webhook (Investment Advisor) · A public landing page with an enquiry form and a floating AI chatbot, both wired to n8n webhooks. A Setup menu in the top nav lets learners paste their own enquiry/chat webhook URLs and admin email (saved in the browser, with a Test button per webhook) — no code editing required.
Activity 6 — Finance API → Telegram (AI Day Trader) · Pulls Twelve Data candles + NewsAPI headlines, then replies with a Buy/Sell/Hold call. A live dashboard shows the chart and quote stats.
Activity 7a — RAG Chatbot: Upload a PDF, Ask in Telegram · A web page extracts text from a PDF (an IT-Support FAQ) and uploads it to n8n, which embeds it into an in-memory vector store with Google Gemini. A Telegram bot then answers only from that document.
Activity 7b — Customer-Support RAG Agent (Cook & Bake Academy) · Ingest 20 course brochures from Google Drive into a vector database, then a CX Agent answers website-chat questions about course duration, fees, and schedule — grounded in the brochures. Shown across three vector stores: Supabase (pgvector), Pinecone, and Qdrant.
Activity 8 — HR Service Portal (Security & Guardrails) · One portal backed by three workflows: human-in-the-loop leave approval (8a), a live leave-balance dashboard (8b), and an AI chatbot wrapped in input/output guardrails (8c).
This repository contains the complete, working lab materials for the WSQ Agentic AI Automation with n8n course (TGS-2023035977) by Tertiary Courses / Tertiary Infotech. Each activity is a self-contained, importable n8n workflow — several paired with a polished HTML front end — that builds progressively from basic automation to a full Retrieval-Augmented Generation (RAG) agent grounded in a vector database.
| # | Activity | Concepts |
|---|---|---|
| 1 | Flyer with QR Code | Form Trigger → Gmail, expressions, QR-code generation |
| 2 | Capture Data in a Data Table | n8n Data Tables, storing submissions |
| 3a / 3b | Conditional Response | IF-node branching → Data Table (3a) / Google Sheets persistence (3b) |
| 4a / 4b | Telegram AI Agent | Telegram Trigger, AI Agent, memory, Data Table tool |
| 5 | Website Chatbot (Investment Advisor) | Webhook trigger, CORS, Respond to Webhook, branded front end |
| 6 | Finance API → Telegram (Day Trader) | HTTP Request, Twelve Data + NewsAPI, multi-timeframe analysis |
| 7a | RAG Chatbot (PDF → Telegram) | PDF upload, embeddings (Gemini), in-memory vector store, retrieve-as-tool |
| 7b | Customer-Support RAG Agent | Google Drive ingestion, vector databases — Supabase, Pinecone (Gemini embeddings), Qdrant |
| 8a / 8b / 8c | HR Service Portal (Guardrails) | Human-in-the-loop approval, live dashboard, pre/post LLM guardrails |
| Capstone | Mini Capstone | End-to-end build (Issue Reporting: form + image → Postgres + gallery) |
📖 Full walkthrough: see LEARNER-GUIDE.md for detailed, click-by-click instructions (with workflow diagrams) for every activity. Slides, the Learner Guide and the Lesson Plan are in
courseware/.
| Category | Technology |
|---|---|
| Automation Platform | n8n (cloud trial or local Docker; workflows, triggers, Data Tables) |
| LLM | OpenAI (chat + text-embedding-3-small) and Google Gemini (chat + gemini-embedding-001) |
| Agent Framework | n8n LangChain nodes (AI Agent, Memory, Vector Store, Tools) |
| Vector Databases | In-memory store · Supabase (pgvector) · Pinecone · Qdrant |
| Chat / Messaging | Telegram (Bot trigger + send) |
| APIs & Data | Twelve Data + NewsAPI (HTTP Request), Google Drive |
| Email / Storage | Gmail (OAuth2), Google Sheets |
| Front End | Vanilla HTML / CSS / JavaScript (no build step), PDF.js |
| Courseware | Slides (python-pptx), Learner Guide + Lesson Plan (python-docx) |
DAY 1 — Workflow Automation + AI Agents
Act 1 Form Trigger ─▶ Gmail (flyer + QR code)
Act 2 Form Trigger ─▶ Gmail + Data Table (capture data)
Act 3a Form ─▶ IF ─▶ Data Table / Gmail (conditional)
Act 3b Form ─▶ IF ─▶ Google Sheets / Gmail (persistent)
Act 4a Telegram ─▶ AI Agent (+ memory) ─▶ reply
Act 4b Telegram ─▶ AI Agent + Data Table tool ─▶ reply
DAY 2 — Webhooks · APIs · RAG
Act 5 Website ─▶ Webhook ─▶ AI Agent ─▶ Respond (Investment Advisor)
Act 6 Telegram ─▶ HTTP (Twelve Data + NewsAPI) ─▶ AI Agent ─▶ reply (Day Trader)
Act 7a Web upload ─▶ Embeddings (Gemini) ─▶ Vector Store │ Telegram ─▶ Agent + knowledge_base ─▶ reply
Act 7b Drive ─▶ split ─▶ Embeddings (OpenAI / Gemini) ─▶ Supabase / Pinecone / Qdrant │ Website ─▶ Webhook ─▶ CX Agent ─▶ reply
DAY 3 — Security & Guardrails + Capstone
Act 8a Form ─▶ Manager Approval (Send & Wait) ─▶ IF ─▶ confirm / decline
Act 8b Webhook ─▶ Code ─▶ Respond JSON (leave dashboard)
Act 8c Webhook ─▶ Input guardrail ─▶ AI Agent ─▶ Output guardrail ─▶ Respond / Blocked
Capstone Issue Reporting: Form + image ─▶ Postgres + retrieval API + gallery
TGS-2023035977-Agentic-AI-Automation-with-n8n/
├── LEARNER-GUIDE.md # Full step-by-step lab guide (start here)
├── README.md
├── screenshot.png # Cook & Bake Academy RAG site (Activity 7b)
│
├── labs/ # All hands-on lab activities (one folder each)
│ ├── n8n-installation/ # Docker Compose for self-hosting n8n
│ ├── activity1-flyer-form/ # Act 1: Form → Gmail (+ flyer samples)
│ ├── activity2-data-table/ # Act 2: + Data Table
│ ├── activity3-conditional/ # Act 3a/3b: IF → Data Table / Google Sheets
│ ├── activity4-telegram-agent/ # Act 4a/4b: Telegram AI agent
│ ├── activity5-investment-advisor/ # Act 5: Webhook website chatbot (HTML app)
│ ├── activity6-finance-advisor/ # Act 6: Finance API → Telegram (HTML dashboard)
│ ├── activity7-rag/ # Act 7a/7b: RAG — PDF→Telegram + vector-DB CX agent
│ │ ├── Activity7a-RAG-Telegram.json · Activity7a-upload.html · it-faq.pdf
│ │ ├── Activity7b-{Supabase,Pinecone,Qdrant}-Upload.json · Activity7b-CX-Agent.json
│ │ ├── brochures/ # 20 mock course brochures (.txt)
│ │ └── website/ # Cook & Bake Academy site + RAG chat widget
│ ├── activity8-guardrails/ # Act 8a/8b/8c: HR Service Portal + guardrails
│ └── mini-capstone/issue-tracking/ # Capstone: Form + image → Postgres + gallery
│
└── courseware/ # Course slides, Lesson Plan + Learner Guide
├── Agentic AI Automation with n8n-v46.pptx # 3-day slide deck (+ PDF)
├── LG-Agentic AI Automation with n8n.docx # detailed step-by-step (+ PDF)
└── LP-Agentic AI Automation with n8n.docx # 3-day lesson plan (+ PDF)
- An n8n account (Cloud or self-hosted via
labs/n8n-installation/) - An OpenAI API key and/or a Google Gemini key (embeddings + chat)
- A Telegram bot token (from @BotFather)
- A Gmail account (Activity 1 email / Activity 8 approvals)
- (Activity 6) free Twelve Data + NewsAPI keys
- (Activity 7b) one vector database — Supabase, Pinecone or Qdrant
git clone https://github.com/tertiarycourses/TGS-2023035977-Agentic-AI-Automation-with-n8n.git
cd TGS-2023035977-Agentic-AI-Automation-with-n8n- In n8n: Workflows → Add workflow → ⋯ → Import from File.
- Pick a
.jsonfrom the matchinglabs/activity*/folder. - Re-select your own credentials on each node — imported credential IDs won't match yours.
- Save, then toggle Active / Published.
The pages are pure static HTML — just open them, or serve locally:
cd labs/activity7-rag/website
python3 -m http.server 8000
# then open http://localhost:8000/index.html- Set the webhook / production URL in the page (gear icon or
script.js). - Activity 7a: open
Activity7a-upload.html, paste therag-uploadwebhook URL, dropit-faq.pdf, then chat with your Telegram bot. - Activity 7b: follow
labs/activity7-rag/LEARNER-GUIDE-7b.mdto ingest the brochures into Supabase / Pinecone / Qdrant.
⚠️ CORS: each n8n Webhook node must have Options → Allowed Origins (CORS) =*so the browser page can call it. All workflow exports in this repo already include this.
For complete, click-by-click setup, see LEARNER-GUIDE.md.
Contributions, fixes, and improvements are welcome:
- Fork the repository
- Create a feature branch:
git checkout -b feature/my-improvement - Commit your changes:
git commit -m "Add my improvement" - Push the branch:
git push origin feature/my-improvement - Open a Pull Request
Found a bug or have an idea? Open an issue.
This material is provided for educational use as part of the WSQ course TGS-2023035977. © Tertiary Infotech Pte. Ltd. All rights reserved.
Tertiary Infotech Pte. Ltd. — Tertiary Courses Course: WSQ Agentic AI Automation with n8n (TGS-2023035977)
- n8n — the workflow automation platform
- OpenAI & Google Gemini — chat & embedding models
- Supabase, Pinecone, Qdrant — vector databases
- Course trainers and learners of TGS-2023035977
⭐ If this helped you learn agentic automation, star the repo!
Powered by Tertiary Infotech Academy Pte Ltd