Memuat
Memuat
Want to go deeper?
Join one of our courses or ask us directly on WhatsApp.
Picked from the software, topic, and level of “Claude from Zero to Expert Class #24: Practice: A Customer Service Project for Sunset Coffee”.
Other ways we can help with Claude.
Learn with a mentor
Scheduled AI & Machine Learning courses with live guidance and a certificate.

Midjourney Course: Create Stunning AI Images from Text
★ 4.9

DALL-E Course: Creating Professional AI Images from Text
★ 4.9

Stable Diffusion Course: AI Art from Beginner to Professional
★ 4.9
Have it built for you
No time to learn Claude? Our team can do the project for you.

AI Image Creation with Stable Diffusion — Unlimited Custom Visual Production
★ 4.9

AI Image & Video Creation with Leonardo AI — All-in-One Visuals for Business
★ 4.9

AI Image Creation with Midjourney — Cinematic AI Art for Business
★ 4.9
Get notified about new Claude classes
At most one email a week. Unfollow anytime with one click.
The full narration of this video. The current sentence is highlighted while the video plays; click any sentence to jump there.
Hi, and welcome to the practice episode of the Claude Projects session. Today we build one complete Project, from zero until it's ready to use. We'll create a customer service Project for Sunset Coffee: plan it, build it, test it with real WhatsApp messages, handle hard cases, then make it part of Dimas's daily routine.
Good practice starts with a plan, not with buttons. Laras and Dimas list the questions that come in most often. The most common are about prices and shipping. Second, ordering: how to order, opening hours, and whether you can order for tomorrow. Third, complaints: spilled coffee, a late delivery, or a taste that's different from usual. Fourth, large orders for office events or family celebrations. Orders like these always need Laras's confirmation, because they involve stock and special prices. These four kinds of questions will become the content of the knowledge files. Try the same thing for your business: open the last month of customer chats, then group the questions. Next, the workflow. Messages arrive via WhatsApp or Instagram DM as usual. Dimas copies the message text into the Project, without the customer's phone number or full address. Claude prepares a reply draft based on the instructions and knowledge files. Dimas checks the draft, especially the numbers and the tone. Only then does Dimas send it from the shop's WhatsApp. Claude doesn't reply to customers automatically; a person stays in control. Full automation is an advanced topic near the end of the class.
Now let's build it. Laras creates a new Project called Sunset Coffee CS, with a description saying Dimas uses it every day from eight in the morning to nine at night. She clicks Edit and writes concise instructions: the admin assistant role, a friendly style that greets customers warmly, limits to answer only from the files and never promise discounts, and a reply draft format. The business context is already in the files, so the instructions stay short. Notice one addition to the format: an admin check tag in square brackets. This tag lets Dimas see at a glance which drafts must not be sent before asking Laras. Next, the knowledge files. First, the December menu and prices, complete with shipping and opening hours. Second, the tidied-up FAQ, covering ordering, delivery, and payment. Third, the SOP for complaints and large orders. For complaints: apologize, ask for a photo, then forward it to Laras. For orders above twenty cups: confirm with Laras first. Fourth, the best sample replies as a style reference. Four files, one concise set of instructions, and the customer service Project is ready to test.
Time to test with real messages. This morning Sunset Coffee's WhatsApp receives a message from Rina, who orders three palm sugar iced coffee cups and asks for the total with shipping. The second message is from Budi: two gift hampers as presents, and he wants to confirm shipping is free. The third message is from Sari, asking whether cold brew can be sent to Cirebon. Dimas copies all three message texts into the Project, without names or phone numbers. These three messages represent three situations: a simple calculation, the free shipping rule, and a request outside the service area. A mix like this is perfect for a first test. Dimas pastes all three into one message, numbered one to three, and asks for a draft for each. Claude writes three drafts at once. For Rina: three times eighteen thousand is fifty-four thousand, plus ten thousand shipping, sixty-four thousand in total. For Budi: three hundred thousand, free shipping. For Sari: a polite decline. Before sending, Dimas checks the numbers with the calculator on his phone. Everything matches. This habit of checking numbers must never go away, no matter how smart Claude is.
Now for the hard cases. Joko is angry because his order still hasn't arrived after two hours, and he wants his money back. Claude doesn't panic along with him. It apologizes, promises to check the courier's location, and forwards the refund question to Laras. At the end there's an admin check tag, so Dimas knows this draft must be confirmed first. The second case: an office wants to order fifty cups for a Friday meeting. Because the SOP in the knowledge files says orders above twenty cups must be confirmed by Laras, Claude doesn't agree right away. It thanks the customer, promises an update, and adds the admin check tag. The third case arrives at night. Putri sends a message at ten thirty, asking whether the shop is still open and wanting to order one liter of palm sugar iced coffee. Claude honestly answers that the shop is closed, gives the opening hours from the file, then offers delivery tomorrow morning, complete with the price and shipping. The customer doesn't feel ignored. Laras notes the results of every hard case test. The angry complaint got an empathetic reply, without defensiveness, and was forwarded to Laras. The large order wasn't accepted right away and was always flagged for an admin check. The message that arrived at ten thirty at night got an honest answer that the shop was closed, plus an offer to order for tomorrow morning. Questions that aren't in the files never got a made-up answer. And most importantly, Claude never promised a discount or compensation on its own. Every limit in the instructions was followed, so this Project passes the test.
Here's the result on WhatsApp. Rina's message arrived at twelve minutes past eight. Three minutes later, Dimas has already replied with the draft he checked. Usually he'd have to open the price list and calculate it himself. Rina is happy because the reply is fast and clear. Speed like this really shows at busy times, when dozens of messages arrive at once. To keep this Project safe for daily use, Dimas follows four rules. First, always read the draft before sending. Claude is the assistant; Dimas is the one responsible. Second, check every number: unit price, quantity, shipping, and total. Third, don't paste personal data. Customers' phone numbers and addresses don't need to go into the Project to create a draft. Fourth, note new questions that don't have an answer yet. Once a week, Laras adds them to the FAQ file so the Project keeps getting smarter. The screens in this video are simulations, and menu names may differ slightly because Claude is updated often. The pattern stays the same for any business, from a bakery to a repair service. Finally, a good Project keeps being cared for. Every Monday morning, Laras and Dimas sit down for ten minutes. First, they gather notes: new questions and drafts that missed the mark during the week. Second, they update the files: adding answers to the FAQ, replacing the menu, or completing the SOP. Third, they fix the instructions if some behavior of Claude needs to change, just one sentence for each problem. Fourth, they test again with that week's customer messages. Ten minutes a week keeps the Project accurate and more and more useful.
We've built a Sunset Coffee customer service Project, from the plan, instructions, and four files, through real message tests and hard cases, to daily rules. In the next episode we learn to share a Project with a team, so Laras, Dimas, and other members work with the same rules and knowledge. See you there!
Try these on your own after watching to lock in what you learned.
Type a question or keyword; we find the part of this episode that explains it and jump the video there.
Notes are saved with the current video time and are only visible to you.
Stuck on a step? Ask here; the instructor and other learners can answer.
No questions yet.

Cerdas Inovasi Digital Team
Software developers, BIM & 3D engineers, and AI practitioners
We build software, BIM models, games and AI automations for clients every day, and turn that hands-on work into step-by-step classes: every step is shown on screen and comes with a PDF guide you can keep.
See our clients →