Meet Juno

Velo Supply sells bike parts online. Three hundred support messages arrive every day, and two people answer all of them. Where's my order? Can I return this? Is this chain in stock? Each answer means opening three systems, copying numbers, and typing a reply. They're exhausted. So the founder gets them a chatbot.

• Velo Supply sells bike parts
• Three hundred messages a day
• Two tired people

What a chatbot can do

Here's the chatbot, in its simplest form. We send it a customer's message. Run it. It's polite, and it sounds helpful. But look closely. Did it check the order? Did it refund anything? No. It can't. It only produces words. It has no hands. The customer still has to wait for a human.

from google import genai

client = genai.Client()

reply = client.models.generate_content(
    model="gemini-flash-latest",
    contents="Order A17 arrived damaged. Can you refund it? Answer in two sentences.")

print(reply.text)

I am very sorry to hear that Order A17 arrived damaged. As an AI, I cannot access order systems or process transactions, so please contact the seller's customer support directly with photos of the damage to receive your refund.

Chatbot versus agent

That's the gap this course closes. A chatbot talks. An agent acts. It reads the order, checks the rules, decides what to do, does it, and tells you what happened. And that's the difference between a tool that saves a few seconds, and one that does the job.

What you will build

By the end of this course, you'll have built Juno, a complete support agent. It reads messages, uses real tools, remembers customers, asks a human before big refunds, resists attacks, and gets tested like real software. Every lesson adds one working piece.

Summary