Assistant Configuration FAQ
The Assistant Configuration system provides a unified approach to customizing your DataQI assistants. This FAQ covers common questions about configuring assistant behavior, tone, and capabilities.
General Configuration
Question: Where can I configure my assistant settings?
Answer: You can configure assistants in two places: during the assistant creation process and on the Manage Assistant page for existing assistants.
Question: Can I change my assistant’s configuration after it’s created?
Answer: Yes, you can modify most settings on the Manage Assistant page. Changes take effect immediately for new conversations.
Question: Do configuration changes affect existing conversations?
Answer: No, configuration changes only apply to new conversations. Existing chat histories remain unchanged.
Assistant Name and Description
Question: How important is the assistant description?
Answer: The description is very important as it helps the AI understand the assistant’s role and context, leading to more accurate and relevant responses.
Question: Can I use the same name for multiple assistants?
Answer: While technically possible, it’s recommended to use unique, descriptive names to avoid confusion and make assistants easier to identify.
Question: How long should the description be?
Answer: Aim for 2-3 sentences that clearly explain the assistant’s purpose, target audience, and key capabilities.
Instructions and Behavior
Question: What types of instructions work best?
Answer: Specific, actionable instructions work best. For example: “Always cite sources” or “If asked about pricing, direct users to contact sales.”
Question: Can I add multiple instructions?
Answer: Yes, you can add multiple instructions. Separate them clearly and ensure they don’t conflict with each other.
Question: How do I test if my instructions are working?
Answer: Ask sample questions that should trigger your instructions and observe how the assistant responds. Adjust instructions based on the results.
Tone of Voice
Question: What’s the difference between the tone options?
Answer:
- Professional: Formal, business-appropriate language
- Friendly: Conversational, approachable style
- Technical: Precise, detailed responses with technical accuracy
- Concise: Brief, to-the-point answers
Question: Can I customize the tone beyond the preset options?
Answer: Currently, you can choose from the preset tone options. Custom tone instructions can be added in the Instructions field.
Question: Should all assistants in my organization use the same tone?
Answer: Not necessarily. Different departments or use cases may benefit from different tones. Choose what works best for each assistant’s specific purpose.
AI Model Selection
Question: How do I choose the right AI model?
Answer: Consider your use case: use faster models for simple tasks and more powerful models for complex analysis or high-stakes content.
Question: Can I change the AI model after creating an assistant?
Answer: Yes, you can change the AI model on the Manage Assistant page, but consider the impact on response quality and speed.
Question: What’s the difference between the available models?
Answer: Models vary in speed, accuracy, and capabilities. Your administrator can provide details about available models and their characteristics.
Troubleshooting
Question: My assistant isn’t following my instructions. What should I do?
Answer: Try making your instructions more specific and clear. Test with sample questions and refine based on the responses you receive.
Question: The tone doesn’t seem right. How can I fix it?
Answer: Try a different tone option or add specific tone instructions in the Instructions field. Test with sample conversations to verify the change.
Question: Can I see how my configuration affects the assistant’s behavior?
Answer: Yes, test your assistant with sample questions after making configuration changes to see how they affect responses.
Best Practices
Question: How often should I review my assistant configurations?
Answer: Review configurations regularly, especially after receiving user feedback or when the assistant’s purpose changes.
Question: Should I configure all assistants the same way?
Answer: No, tailor each assistant’s configuration to its specific purpose, audience, and use case for optimal results.
Question: How can I ensure my configurations are effective?
Answer: Test thoroughly with sample questions, gather user feedback, and iterate based on real-world usage patterns.