✦ What you'll build: A conversational agent that helps customers order from a bakery menu, handles the full ordering flow, collects delivery details, and confirms orders — all in a natural, friendly tone.
Step 1: Prepare Your Context
Before building the agent, you need to create the knowledge it will use — in this case, a menu for Penny's Little Bakery.
Create a Menu
In a Channel, mention the @Product Agent and ask it to help you build a menu for Penny's Little Bakery. Include breakfast items, lunch options, drinks, and prices.
Share the Menu to Your Space
Once Product Agent creates the menu, ask it to share this menu on the Space so it's accessible to other agents and conversations. This makes the menu available as context for your ordering agent.
💡 Pro tip: Sharing files or content to your Space makes them available as context to all agents in that Space — this is how your custom agent will "know" the menu without you having to paste it every time.
Step 2: Start Building with Agent Builder
Now you're ready to create the ordering agent. Open a new conversation (or continue in the same Channel) and mention @Agent Builder.
Tell Agent Builder to Use the Menu
Ask Agent Builder to get the menu you just created and use it as context for building the new agent. Agent Builder will retrieve the menu from your Space files.
Agent Builder Starts Asking Questions
Agent Builder will ask clarifying questions to understand what you want to build. The first question is usually: "What should we name this agent?"
Step 3: Define Agent Behavior
This is the most important step — describing exactly how your agent should behave. Agent Builder will ask two key questions:
QUESTION 1: WHAT'S THE AGENT'S NAME?
Agent Name: "Penny's Little Bakery Agent"
Agent Name: "Penny's Little Bakery Agent"
Agent Name: "Penny's Little Bakery Agent"
QUESTION 2: HOW SHOULD THE AGENT BEHAVE?
Describe the conversation flow in natural language. Here's an example of what to provide:
Flow of Conversation:
1. Greet the customer warmly and introduce Penny's Little Bakery 2. Ask what type of food they're interested in (breakfast, lunch, drinks) 3. Show relevant menu items based on their interest 4. Take their order — allow them to add multiple items 5. Confirm each item as they order it 6. When they're done ordering, show the order summary with: - Item list - Subtotal - Delivery fee ($5) - Total 7. Ask for delivery information: - Street address - City - Phone number 8. Ask for payment method (cash or card) 9. Confirm the complete order with: - Order number (generate a random 6-digit number) - Estimated delivery time (simulate 30-45 minutes)Tone: The agent should be polite, friendly, and conversational. Make the customer feel welcome and help them find what they want.
Flow of Conversation:
1. Greet the customer warmly and introduce Penny's Little Bakery 2. Ask what type of food they're interested in (breakfast, lunch, drinks) 3. Show relevant menu items based on their interest 4. Take their order — allow them to add multiple items 5. Confirm each item as they order it 6. When they're done ordering, show the order summary with: - Item list - Subtotal - Delivery fee ($5) - Total 7. Ask for delivery information: - Street address - City - Phone number 8. Ask for payment method (cash or card) 9. Confirm the complete order with: - Order number (generate a random 6-digit number) - Estimated delivery time (simulate 30-45 minutes)Tone: The agent should be polite, friendly, and conversational. Make the customer feel welcome and help them find what they want.
Flow of Conversation:
1. Greet the customer warmly and introduce Penny's Little Bakery 2. Ask what type of food they're interested in (breakfast, lunch, drinks) 3. Show relevant menu items based on their interest 4. Take their order — allow them to add multiple items 5. Confirm each item as they order it 6. When they're done ordering, show the order summary with: - Item list - Subtotal - Delivery fee ($5) - Total 7. Ask for delivery information: - Street address - City - Phone number 8. Ask for payment method (cash or card) 9. Confirm the complete order with: - Order number (generate a random 6-digit number) - Estimated delivery time (simulate 30-45 minutes)Tone: The agent should be polite, friendly, and conversational. Make the customer feel welcome and help them find what they want.
✦ Key principle: Describe the agent's behaviour in natural language — just like you're training a new employee. The more detailed your flow, the more accurately Agent Builder can implement it.
Step 4: Agent Builder Creates Your Agent
Once you've provided the behavior description, Agent Builder gets to work:
Agent Builder creates a plan for building all necessary components
It implements the agent logic based on your flow description
It automatically generates a test web application— a live chat interface where you can interact with your agent
Agent Builder provides a URLto the test application
⚠️ Important for Hackathon Participants: Currently, LangGraph agents built with Agent Builder need to connect to your own LLM provider. You'll need to provide your own API keys (OpenAI, Anthropic, AWS Bedrock, etc.) in the agent's configuration. Complete does not currently provide system LLM endpoints for LangGraph agents during the hackathon.
🔮 Coming Soon: Direct access to Complete's LLMs for LangGraph agents is in development. If you need this for your project, reach out to the Complete team in the hackathon Discord.
You can instruct Agent Builder to use any LLM provider you have access to. For example: "Use AWS Bedrock with Claude" or "Use OpenAI GPT-4 with my API key."
Step 5: Test Your Agent
Click the URL Agent Builder provides. You'll see a chat interface where you can test your agent in real-time.
🧪
Example Test Conversation
You: Hi, what breakfast options do you have?
Agent: Good morning! For breakfast we have Croissants, Quiche Lorraine, Fruit Parfait, and Breakfast Sandwich. What would you like?
You: I'll take two slices of Quiche Lorraine
Agent: Great choice! Two slices of Quiche Lorraine added to your order. What else can I get you?
You: A coffee please — black
Agent: One Café Espresso (black) added! Anything else, or are you ready for delivery details?
You: That's it, let's do delivery
Agent: Perfect! What's your street address?
🧪
Example Test Conversation
You: Hi, what breakfast options do you have?
Agent: Good morning! For breakfast we have Croissants, Quiche Lorraine, Fruit Parfait, and Breakfast Sandwich. What would you like?
You: I'll take two slices of Quiche Lorraine
Agent: Great choice! Two slices of Quiche Lorraine added to your order. What else can I get you?
You: A coffee please — black
Agent: One Café Espresso (black) added! Anything else, or are you ready for delivery details?
You: That's it, let's do delivery
Agent: Perfect! What's your street address?
🧪
Example Test Conversation
You: Hi, what breakfast options do you have?
Agent: Good morning! For breakfast we have Croissants, Quiche Lorraine, Fruit Parfait, and Breakfast Sandwich. What would you like?
You: I'll take two slices of Quiche Lorraine
Agent: Great choice! Two slices of Quiche Lorraine added to your order. What else can I get you?
You: A coffee please — black
Agent: One Café Espresso (black) added! Anything else, or are you ready for delivery details?
You: That's it, let's do delivery
Agent: Perfect! What's your street address?
Notice how the agent:
Follows the exact flow you described
Understands natural language input ("two slices", "black coffee")
Maintains context throughout the conversation
Uses the polite, friendly tone you specified
Step 6: Refine the Agent
If something doesn't work as expected during testing, simply tell Agent Builder what needs to change:
"The agent should also offer recommendations when showing menu items. For example, suggest popular combinations like 'Many customers love the Quiche Lorraine with a Cappuccino!'"
"The agent should also offer recommendations when showing menu items. For example, suggest popular combinations like 'Many customers love the Quiche Lorraine with a Cappuccino!'"
"The agent should also offer recommendations when showing menu items. For example, suggest popular combinations like 'Many customers love the Quiche Lorraine with a Cappuccino!'"
Agent Builder will:
Update the agent with your changes
Regenerate the test application
Provide a new URL to test the updated version
🔄 Iteration is key: Agent building is an iterative process. Test, refine, test again. Keep going until your agent behaves exactly how you need it to.
Step 7: Complete Order Flow Test
Continue testing the full flow to make sure everything works:
✅
Full Order Completion
Agent: What's your street address?
You: 123 Jerry Street
Agent: And which city?
You: Baker County
Agent: Great! And your phone number?
You: 555-0123
Agent: How would you like to pay — cash or card?
You: Cash
Agent: Perfect! Here's your order confirmation:
Order #847392
• 2x Quiche Lorraine - $14.00
• 1x Café Espresso - $3.50
Subtotal: $17.50
Delivery: $5.00
Total: $22.50
Delivering to: 123 Jerry Street, Baker County
Phone: 555-0123
Payment: Cash on delivery
Estimated arrival: 35-40 minutes
Thank you for choosing Penny's Little Bakery! 🥐
✅
Full Order Completion
Agent: What's your street address?
You: 123 Jerry Street
Agent: And which city?
You: Baker County
Agent: Great! And your phone number?
You: 555-0123
Agent: How would you like to pay — cash or card?
You: Cash
Agent: Perfect! Here's your order confirmation:
Order #847392
• 2x Quiche Lorraine - $14.00
• 1x Café Espresso - $3.50
Subtotal: $17.50
Delivery: $5.00
Total: $22.50
Delivering to: 123 Jerry Street, Baker County
Phone: 555-0123
Payment: Cash on delivery
Estimated arrival: 35-40 minutes
Thank you for choosing Penny's Little Bakery! 🥐
✅
Full Order Completion
Agent: What's your street address?
You: 123 Jerry Street
Agent: And which city?
You: Baker County
Agent: Great! And your phone number?
You: 555-0123
Agent: How would you like to pay — cash or card?
You: Cash
Agent: Perfect! Here's your order confirmation:
Order #847392
• 2x Quiche Lorraine - $14.00
• 1x Café Espresso - $3.50
Subtotal: $17.50
Delivery: $5.00
Total: $22.50
Delivering to: 123 Jerry Street, Baker County
Phone: 555-0123
Payment: Cash on delivery
Estimated arrival: 35-40 minutes
Thank you for choosing Penny's Little Bakery! 🥐
The agent successfully:
Collected all delivery information step-by-step
Calculated the subtotal, delivery fee, and total correctly
Generated a random order number
Provided a simulated delivery time
Confirmed everything in a clear, organized format
Step 8: Deploy to Your Workspace
Once you're satisfied with how the agent performs, tell Agent Builder you're ready to deploy. Agent Builder will:
Finalize the agent configuration
Add it to your workspace's Agent Library
Make it available to add to any Space or Channel
Your team can now use "Penny's Little Bakery Agent" anywhere in your workspace — in customer-facing channels, support threads, or even embedded in other workflows.
Key Takeaways
Prepare context first — create and share any menus, docs, or knowledge your agent needs
Describe behavior in natural language — write out the conversation flow like you're training a person
Be specific about tone — friendly, formal, concise, detailed — the clearer you are, the better
Test thoroughly — Agent Builder generates a live test app so you can try every scenario
Iterate freely — ask Agent Builder to make changes until the agent behaves perfectly
Deploy when ready — your custom agent becomes available to your whole team
🎉 Congratulations! You've built a fully functional conversational agent from scratch using only natural language. No code, no APIs to manage, no infrastructure to set up, just conversation with Agent Builder.
