Playground
The Playground is a testing environment where you can interact with your AI agent to evaluate its responses to various inputs. It enables you to fine-tune the agent's behavior, ensuring it aligns with your needs before deploying it to your website or other platforms.
Model Selection:
- Choose an appropriate AI model for your chatbot. For instance, advanced models (like GPT-4 variants) are better for handling nuanced conversations.
- The model selected impacts the chatbot's capabilities and response quality.
Temperature:
Adjust the slider to control creativity in responses:
- Lower values (e.g., 0.1) make the chatbot respond with more predictable and focused answers.
- Higher values (e.g., 0.7) make responses more varied and creative.
Example: For legal guidance, a low temperature ensures consistent and accurate information.
AI Actions: (Coming Soon)
Define specific actions your chatbot can perform, like retrieving data, responding to FAQs, or triggering workflows (Online booking, shopping, etc).
Example: Add an action for "Generate a PDF report" or "Redirect to support."
System Prompt:
Write instructions that define the chatbot’s purpose and tone.
Example: If it’s a legal chatbot, your prompt could be:
"You are a legal assistant providing accurate legal advice based on local laws."
Instructions:
Further refine the chatbot’s role and persona.
Example: If your chatbot is customer support-focused, describe the tone, purpose, and limitations, such as:
"You are a polite and helpful customer support agent. Provide precise solutions and escalate issues when needed."