"Can I really build a chatbot without coding? In 30 minutes?"
Yes. And by the end of this article, you will have done it.
Microsoft Copilot Studio (formerly Power Virtual Agents) is Microsoft's low-code platform
for building AI agents. It handles the Natural Language Understanding, the conversation flow,
the knowledge retrieval, and the deployment pipeline — you just describe what the bot should
do. The platform does the rest.
This guide is for someone who has never built a chatbot before. We will build
HelpBot — a simple Teams bot that answers common employee questions about
office policies using a SharePoint document as its knowledge source.
✅ What You Need Before You Start
- A Microsoft 365 account with Power Platform access (work or school account)
- Access to make.microsoft.com (the Power Platform maker portal)
- A SharePoint page or a PDF document you want the bot to answer questions about
(even a one-page Word doc uploaded to SharePoint works)
- 30 minutes of uninterrupted time
No Azure subscription needed. No coding. No bot framework.
🚀 Step 1 — Create Your Agent (5 minutes)
1
Open the maker portal
Go to make.microsoft.com in your browser.
Sign in with your Microsoft 365 account. You will land on the Power Platform home page.
↓
2
Find Copilot Studio
In the left navigation, scroll down and click
Copilot Studio. If you don't see it, search for it using the
search bar at the top of the left panel.
↓
3
Create a new copilot
Click + Create in the top left →
Select New copilot → Fill in:
Name: HelpBot · Language: English · Description: "Answers employee
questions about office policies." · Click Create.
↓
✓
Your agent canvas opens
You now have a blank agent. On the left you'll see:
Topics, Knowledge, Actions, Analytics, and Settings. We'll use Topics and Knowledge
in this guide.
📚 Step 2 — Add a Knowledge Source (5 minutes)
A knowledge source is a document, SharePoint page, or website the bot reads to answer
questions. Without one, the bot can only use its general AI training — which won't know
about your company's specific policies.
1
Go to the Knowledge tab
In the left panel click Knowledge →
then click + Add knowledge.
↓
2
Choose a source type
Options include: SharePoint, public website, uploaded
file, or Dataverse. For this guide, choose Public website and enter
your company intranet URL — or choose Upload file and upload a PDF
of your office policy.
↓
✓
Source added and indexing
Copilot Studio indexes the content. This takes
1–3 minutes. A green checkmark appears when indexing is complete. The bot can now
retrieve information from this document when users ask questions.
💡 Enable Generative Answers
Go to Settings → Generative AI → toggle on
Generative answers. With this on, if a user asks something not covered
by a specific topic, the bot answers using your knowledge source automatically.
This is what makes the 30-minute bot actually useful.
💬 Step 3 — Build Your First Custom Topic (10 minutes)
A Topic is a specific conversation the bot knows how to handle — activated when a user
says certain phrases. Let's build one for the most common question: "What are the office hours?"
1
Open Topics → Add a topic
Click Topics in the left panel →
click + Add a topic → choose From blank.
↓
2
Name the topic and add trigger phrases
Name: Office Hours.
Under Phrases, add these trigger phrases one by one:
"What are the office hours", "When is the office open", "Office timing",
"What time does the office open", "Working hours".
The more variants you add, the better the NLU matching.
↓
3
Add a Message node
Click the + button below the trigger →
choose Send a message → Type your answer:
"Our office is open Monday to Friday, 9:00 AM to 6:30 PM IST.
The building is closed on public holidays. For WFH policy, ask me about 'work from home'."
↓
✓
Save the topic
Click Save in the top right.
Your first topic is live inside the bot. When a user asks any of the trigger phrases,
this exact message will be sent back.
🧪 Step 4 — Test the Bot (5 minutes)
Before publishing, always test in the built-in Test Panel.
1
Open the Test Panel
Click Test your copilot in the
bottom-left of the screen. A chat panel slides open on the right.
↓
2
Test your topic
Type: "When does the office open?"
The bot should respond with your Office Hours message.
Try a variant: "office timing" — it should still trigger correctly.
Enable Debug mode (toggle at the top of the test panel) to see
exactly which topic fired and what the NLU confidence score was.
↓
3
Test the knowledge source
Now ask something from your uploaded document that
you did NOT create a topic for: e.g. "What is the leave encashment policy?"
The bot should answer from the document via generative answers — not from a topic.
This is the power of knowledge sources combined with generative AI.
🚀 Step 5 — Publish to Microsoft Teams (5 minutes)
1
Click Publish
In the top right of Copilot Studio, click
Publish → confirm in the dialog. Publishing takes about 1 minute.
↓
2
Go to Channels → Microsoft Teams
In the left panel click Settings →
Channels → click Microsoft Teams →
click Turn on Teams.
↓
✓
Open in Teams
Click Open the bot — your browser
opens Microsoft Teams with HelpBot as a new chat. Type "What are the office hours?"
in Teams. Your bot responds. It is live.
⚠️ Sharing with your team
To share with others, you can send them the Teams deep link from the Channels page,
or ask your Teams admin to publish the bot as a Teams app so it appears in the
Apps section for all users in your organisation.
📌 What You Just Built — and What's Next
In 30 minutes you created an AI bot with:
- A named agent with custom personality via description
- A knowledge source that answers open questions from a document
- A custom topic that handles a specific, common question with a precise answer
- Generative AI fallback for questions without a dedicated topic
- A live deployment in Microsoft Teams
Where to go next: Add more topics for your top
10 most common questions. Connect a Power Automate flow to let the bot actually submit
requests (leave, IT tickets, bookings). Add Adaptive Cards for richer UI.
Read Raushan's full Copilot Studio story-based guide for a deep-dive into every concept.
Go Deeper
Read "Raushan and the Rise of the Smart Agent" — the full story-based guide covering
Topics, Entities, Variables, Power Automate integration, NLU training, testing, and
deployment. Every concept you need for enterprise Copilot Studio development.