Building Your First Agent
Step-by-step guide to creating your first AI agent using our visual builder.
Introduction to the Visual Builder
NabkaAI's Visual Builder lets you create AI agents by dragging and dropping nodes onto a canvas. No coding required! Each node represents a step in your agent's workflow.
What We'll Build
In this tutorial, you'll create a Customer Support Agent that:
- Greets customers warmly
- Understands their questions
- Provides helpful, friendly responses
- Knows when to escalate to a human
Prerequisite: Make sure you have a NabkaAI account. If not, sign up here (it's free!).
1Access the Visual Builder
- a
Log in to your NabkaAI account at nabkaai.web.app/login
- b
Click "Builder" in the navigation menu
- c
Click the "+ Create New Agent" button
Builder Interface Overview
Left Sidebar
Toolbar with node types you can drag onto the canvas
Center Canvas
Where you build your agent by connecting nodes
Right Panel
Configuration options for the selected node
2Add a Trigger Node
Every agent starts with a Trigger Node. This is the entry point that activates when a user sends a message.
How to Add:
- 1. Find "Trigger" in the left toolbar
- 2. Click and drag it onto the canvas
- 3. The Trigger node appears with a green "Start" label
Tip: You can only have one Trigger node per agent. It automatically receives the user's input message.
3Add a Prompt Node (System Prompt)
The Prompt Node contains instructions that define your agent's personality and behavior. This is the most important part!
How to Add:
- 1. Drag a "Prompt" node from the toolbar
- 2. Place it to the right of your Trigger node
- 3. Click the node to select it
- 4. In the right panel, enter your system prompt
Example System Prompt:
You are a friendly and helpful customer support agent for TechCorp, a software company. Your role: - Greet customers warmly - Answer questions about our products - Help troubleshoot common issues - Be patient and understanding Rules: - Always be polite and professional - If you don't know the answer, say so honestly - For billing issues, tell users to email billing@techcorp.com - For urgent issues, suggest calling our hotline: 1-800-TECH Response style: - Keep responses concise (2-3 sentences when possible) - Use simple, clear language - Add relevant emojis sparingly to seem friendly 😊
Pro Tips for Great Prompts:
- • Be specific - Vague prompts lead to inconsistent behavior
- • Include examples - Show the agent how to respond
- • Set boundaries - Define what the agent should NOT do
- • Define edge cases - What should happen in unusual situations?
4Add an LLM (AI) Node
The LLM Node is where the AI magic happens. It takes the prompt and user message, processes them, and generates a response.
How to Add:
- 1. Drag an "LLM Call" node from the toolbar
- 2. Place it to the right of your Prompt node
- 3. Click to select and configure in the right panel
LLM Configuration Options:
Model
Choose the AI model. We use Llama 3.1 via Groq for fast responses.
Temperature
Controls creativity. Lower (0.1-0.3) = more consistent. Higher (0.7-1.0) = more creative. For support, use 0.3.
Max Tokens
Maximum response length. 256-512 is good for chat. Set to 512.
5Add an Output Node
The Output Node sends the AI's response back to the user. Every agent workflow must end with an Output node.
How to Add:
- 1. Drag an "Output" node from the toolbar
- 2. Place it to the right of your LLM node
- 3. This node automatically displays the LLM response
6Connect the Nodes
Now we need to connect the nodes so data flows from one to the next.
How to Connect:
- 1. Hover over the right edge of the Trigger node until you see a handle
- 2. Click and drag from the handle to the left edge of the Prompt node
- 3. Release to create a connection line
- 4. Repeat: Connect Prompt → LLM → Output
Your Final Flow:
7Name and Save Your Agent
Final Steps:
- 1. Click the agent name field in the header (usually "Untitled Agent")
- 2. Enter a name: "TechCorp Support Agent"
- 3. Add a description: "Friendly customer support for TechCorp products"
- 4. Click the "Save" button
🎉 Congratulations!
You've built your first AI agent! In the next tutorial, we'll test it to make sure it works as expected.
Key Takeaways
- ✓Every agent needs: Trigger → Prompt → LLM → Output
- ✓The system prompt is the most important part—be specific!
- ✓Lower temperature = more consistent, higher = more creative
- ✓Connect nodes by dragging from handles on the edges