Skip to main content
n8nAmbassadorCommunity-run site
← Back to the event page// THE TOPICS

10 challenges, 3 levels, one dataset each

Pick a topic that fits your team, start with the Easy level and upgrade if you have time left. Every topic below ships with a downloadable sample dataset — copy it to your own Drive before you start building.

Dates and due dates in every dataset are calculated against 19 Sep 2026 — treat that as "today."
Vote for your favorite team
01

AI Lead Intake Assistant

Furniture, agencies, training centres

Furniture showrooms, agencies and training centres miss enquiries or reply too late, so warm leads go cold.

  1. Easy~20 min

    A form captures the enquiry → save it to Google Sheets → notify sales.

  2. Medium~40 min

    AI reads the request, classifies the service and its priority (hot / warm / cold), then drafts a tailored first reply.

  3. HardStretch

    Route each lead to the right salesperson by specialty, flag duplicates (same phone or email), and send a reminder when a lead is still untouched after the SLA.

Suggested nodes

n8n Form TriggerGoogle SheetsText ClassifierAI AgentSwitchWaitGmail / Telegram

Normal case

A clear request for a 3-bedroom interior design package.

Edge case

The same customer submits twice under a different name.

Tip. For the SLA demo, set the Wait node to 1–2 minutes instead of hours.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

02

Chat Message to Order

Online shops, retail stores

Shops take orders through free-text chat. Staff re-type every order and often get products or quantities wrong.

  1. Easy~20 min

    An order form → look up the price list → calculate the total and generate an order confirmation.

  2. Medium~40 min

    AI turns a Vietnamese chat message into a structured order (items, quantity, phone, address) and asks the customer when details are missing.

  3. HardStretch

    Check stock, suggest a substitute when an item is sold out, and wait for the seller's approval before recording the order.

Suggested nodes

Webhook / Form TriggerInformation ExtractorGoogle SheetsCodeIfSend and Wait for Response

Normal case

“Cho em 2 áo thun trắng size M, giao 12 Lê Lợi, Q1.”

Edge case

A message with no delivery address, or an item that is out of stock.

Tip. Give the Information Extractor a clear schema: items[], phone, address.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

03

Payment Reminder Autopilot

Distributors, B2B services

Receivables live in Excel. Reminder dates get forgotten and cash comes in late.

  1. Easy~20 min

    Read the receivables sheet → filter invoices that are due → create reminder emails.

  2. Medium~40 min

    Group invoices into due soon / overdue / paid, and write a reminder in the right tone for each group.

  3. HardStretch

    Match receivables against sample bank transactions, mark the matches, send unclear cases to accounting, and stop reminders for invoices already paid.

Suggested nodes

Schedule TriggerGoogle SheetsFilterSwitchMergeCodeGmail

Normal case

An invoice 5 days overdue gets a polite but firm reminder.

Edge case

A customer paid only part of the amount, or used the wrong payment reference.

Tip. Keep money logic in rules and Code. Use AI only for the wording.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

04

After-sales Care Desk

E-commerce, service businesses

Complaints arrive scattered across channels, and nobody knows what to handle first.

  1. Easy~20 min

    Receive feedback → create a ticket ID → log it and notify the responsible team.

  2. Medium~40 min

    AI classifies each case as delivery, return or warranty, detects urgency and drafts a reply.

  3. HardStretch

    Check the return and warranty policy to recommend a resolution. Escalate to a person when the policy doesn't cover the case, and send reminders on overdue tickets.

Suggested nodes

Form Trigger / WebhookText ClassifierAI AgentGoogle SheetsSwitchGmail / Telegram

Normal case

A customer complains that a delivery is 3 days late.

Edge case

An angry customer asks for a refund outside the 7-day return window.

Tip. Paste the short policy into the agent's system message; no vector store needed.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

05

Low-stock Early Warning

Shops, cafés, distributors

Shops and cafés only notice a shortage when a customer tries to buy.

  1. Easy~20 min

    Read the stock sheet → alert on items below their minimum threshold.

  2. Medium~40 min

    Use average daily sales and supplier lead time to estimate the stock-out date and suggest how much to reorder.

  3. HardStretch

    Group the suggestions by supplier → draft purchase requests → a manager approves them before the status is updated.

Suggested nodes

Schedule TriggerGoogle SheetsCodeAggregateSend and Wait for ResponseGmail

Normal case

Coffee beans drop below the reorder level.

Edge case

An item with no sales history or no lead time recorded.

Tip. Days left = stock ÷ average daily sales. Reorder when days left ≤ lead time.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

06

Booking & Reminder Assistant

Salons, spas, training centres

Staff spend hours confirming bookings, and customers forget appointments or get double-booked.

  1. Easy~20 min

    Receive a booking request → save it → send a confirmation.

  2. Medium~40 min

    Check free slots, offer alternative times when there is a clash, and send a reminder before the appointment.

  3. HardStretch

    Allow rescheduling and cancellations. When a slot opens up, invite people on the waitlist and confirm only the first one who accepts.

Suggested nodes

Form TriggerGoogle Sheets / Google CalendarIfWaitGmail / Telegram

Normal case

A customer books an open slot for Saturday at 10:00.

Edge case

Two customers request the same slot at almost the same time.

Tip. Use a status column (held → confirmed) to prevent confirming the same slot twice.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

07

Invoice Capture & Expense Approval

Any SME finance team

Employees send invoices through many channels. Accounting re-types them by hand and details are often missing.

  1. Easy~20 min

    A payment-request form with invoice upload → store the file → notify accounting.

  2. Medium~40 min

    AI extracts the supplier, date and amount from the image or PDF → writes them to the sheet and flags fields to double-check.

  3. HardStretch

    Detect likely duplicates, check spending limits using sample rules, and route each request to the right approver.

Suggested nodes

Form Trigger (file upload)Extract from FileInformation ExtractorGoogle DriveGoogle SheetsSwitchSend and Wait for Response

Normal case

A clear PDF invoice under the auto-approval limit.

Edge case

The same invoice is submitted twice, or the photo is blurry.

Tip. Use a vision-capable model, and flag low-confidence fields rather than guessing.
08

Store Content Assistant

Retail, F&B, local brands

SMEs have no marketing staff and spend hours writing product and promotion posts.

  1. Easy~20 min

    Enter product details → AI writes a Vietnamese post → save it as a draft.

  2. Medium~40 min

    Create versions for Facebook, Zalo and email in the brand voice, and check prices and offers against the input data.

  3. HardStretch

    Add an approve / request-changes loop and scheduling. Only approved content moves to the publishing queue.

Suggested nodes

Form TriggerBasic LLM Chain / AI AgentCodeGoogle SheetsSend and Wait for Response

Normal case

A 20%-off weekend promotion for a new milk tea.

Edge case

The AI invents a discount that isn't in the input, and the check catches it.

Tip. Compare the prices in the AI output with the input in a Code node before approval.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

09

Warranty Intake & Dispatch

Electronics, computer & device shops

Shops selling electronics, computers and devices struggle to track repair requests and assign technicians.

  1. Easy~20 min

    Receive a request → create a warranty ticket → send the customer a tracking code.

  2. Medium~40 min

    Check the purchase date against a sample policy to decide warranty status, then classify the fault type and area.

  3. HardStretch

    Assign a technician by area and skill, send reminders on late jobs, and update the customer on progress.

Suggested nodes

Form TriggerGoogle SheetsDate & TimeText ClassifierSwitchGmail / Telegram

Normal case

A laptop bought 4 months ago that won't power on.

Edge case

The warranty has expired, or the serial number isn't found.

Tip. Warranty end = purchase date + warranty months, calculated with the Date & Time node.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.

10

Owner's End-of-Day Report

Any SME owner

Owners have to open several files to understand revenue, orders, receivables and open issues.

  1. Easy~20 min

    Aggregate today's sales → send revenue, order count and average order value.

  2. Medium~40 min

    Compare with yesterday or the target, and let AI explain the changes in plain Vietnamese.

  3. HardStretch

    Combine revenue, receivables and stock → recommend the top three priorities, each backed by numbers.

Suggested nodes

Schedule TriggerGoogle SheetsAggregateCodeBasic LLM ChainTelegram / Gmail

Normal case

A normal day with revenue 12% above yesterday.

Edge case

A day with zero sales, or a data file that is missing.

Tip. Calculate every number in Code, and let AI only explain them.

Sample dataset

Fictional data only — swap in your own test email or chat channel before you demo.