Rules

Rules are essential guidelines that govern your AI Agent’s behavior. They make sure the Agent stays professional, avoids sensitive or incorrect information, and responds in a tone that matches your standards.

Where Guardrails protect your AI from harmful inputs,
Rules shape how your AI should behave at all times.

Together, they create a controlled, reliable, and trustworthy AI experience.

Why Rules Matter

Rules are important because they help your Agent:

1. Stay Consistent

Your Agent follows the same communication principles every time.

2. Stay Professional

Rules prevent inappropriate language, errors, or unsafe replies.

3. Stay Trustworthy

Preventing misinformation protects your brand and user confidence.

4. Stay Compliant

They help keep conversations aligned with legal, ethical, or company guidelines.

5. Improve User Satisfaction

Rules guide your Agent on how to respond in difficult or emotional situations — calmly, clearly, and helpfully.

How to Set Rules in MagicBlocks

Rules are managed inside the Guardrails section of your Agent.

Here’s how to set them:

1. Go to Agents in your sidebar.
2. Select the Agent you want to edit.
3. Open General → Guardrails.
4. Find the Rules area.
5. Write clear instructions for how your AI should behave.

Your Agent will now follow these rules across all conversations.

What Kind of Rules Should You Add?

Here are common examples:

• “Always stay polite and empathetic.”
• “Never provide legal or medical advice.”
• “Avoid making guarantees about results or timelines.”
• “If you’re unsure, ask the user for clarification instead of guessing.”
• “Never mention internal company information.”

Well-written rules help define your Agent’s boundaries and ensure quality output.

Best Practices

• Keep rules short and specific
• Use simple language the AI can follow
• Focus on behavior, not content structure
• Avoid writing vague rules like “Be helpful”
• Update your rules as your company standards evolve

Related Articles

Guardrails

Advanced Features: Guardrails

FAQs

Q: Are Rules the same as Guardrails?
No.
Guardrails protect the AI from unsafe inputs and harmful content.
Rules shape how your AI behaves during normal conversation.

Q: Do Rules apply everywhere?
Yes — Rules act globally across your Agent.

Q: Do I need Rules if I already added a Persona?
Yes. Persona controls tone.
Rules control behavior and boundaries.

Q: Will Rules override my knowledge or playbook?
If there’s a conflict, Rules take priority to keep your Agent safe and compliant.

Q: How many Rules should I add?
Start with 3–5 clear rules, then grow as needed.