Prompt Engineering with Lamatic

In this guide, we’ll explore how to use Lamatic.ai's built-in Prompt Editor to create modular and testable prompts using multiple roles—System, User, and Assistant.

What You’ll Learn

• Create prompt blocks using roles
• Inject variables into your prompts
• Test and iterate inside Lamatic Studio
• Apply version control strategies for scalable updates

1. Getting Started

1. Open any LLM-enabled flow inside Lamatic.ai Studio.
2. Drag a Generate Text Node or click into an existing one.
3. Click into to launch the editor.

2. Understanding Prompt Roles

You can define prompts under three distinct roles:

• System: Core behavior, tone, and assistant persona.
• User: Input coming from your app or user directly.
• Assistant: (Optional) Past assistant replies to retain tone continuity.

Switch roles using the dropdown in the prompt editor.

System: You are a helpful assistant trained on company knowledge.
User: {{triggerNode_1.output.chatMessage}}

3. Injecting Variables

You can use the top panel to insert dynamic variables, or simply type {{ to open the variable selection menu.

{{topic}}
{{userName}}

Use these to personalize or parameterize prompts.

Example:

System: Welcome {{userName}}, I’ll help you with {{topic}}.

4. Save the Prompt as Template

Click the + icon at the top right of each prompt type section to save it as a prompt template. This allows you to easily reuse it across other flows.

You can view all your saved prompts on the Prompts page in the Studio.

5. Testing & Debugging

Click Test on the LLM node to evaluate output.

• You’ll see the input rendered with variable values.
• The Test Output panel shows final model output.
• The Raw Output panel can help identify unexpected behavior or formatting issues.

6. Versioning Strategy

Each prompt can be versioned by naming conventions like:

help-desk-v1
help-desk-v1.1
help-desk-v2

Use “duplicate template” to fork and evolve prompts safely. Deprecated ones can be marked -deprecated to avoid accidental reuse.

Summary

By using Lamatic's Prompt Editor:

• You modularize LLM behavior cleanly
• Keep system/user logic separated
• Enable reusability across flows
• Streamline debugging and controlled updates