System Messages & Memory

In This Section, You Will Learn:

• • What a System Message is (with definition)
• • How to give your bot a personality and purpose
• • What Memory does in AI agents
• • Configuring short-term and long-term memory
• • Making your bot remember conversations
• • Industry-specific bot examples

What is a System Message? (Definition)

• System Message = The instructions that tell the AI who it is, what it does, and how to behave.
• Think of it as: The job description and training manual for your AI employee.
• Without System Message:
• • Bot is a generic AI
• • Talks about anything
• • No specific purpose
• With System Message:
• • Bot is a specialized worker
• • Knows its role and limits
• • Provides consistent, focused responses

Writing Effective System Messages

• Good System Message Structure:
• 1. Identity: 'You are [name/role] for [company]'
• 2. Purpose: 'Your job is to [main task]'
• 3. Knowledge: 'You know about [topics]'
• 4. Behavior: 'Always be [tone/style]'
• 5. Limitations: 'You cannot [what to avoid]'
• 6. Actions: 'When user asks X, do Y'
• Example for a Salon:
• 'You are Sarah, the virtual assistant for Glamour Salon. Help customers book appointments, answer pricing questions, and provide service information. Be friendly and professional. Our services include haircuts ($30), coloring ($80), and styling ($50). If asked about medical advice, politely decline.'

What is Memory? (Definition)

• Memory = The ability for the bot to remember previous messages in the conversation.
• Without Memory:
• • Bot treats every message as new
• • Can't follow up on previous topics
• • User: 'How much is a haircut?' -> Bot: '$30'
• • User: 'Book one for tomorrow' -> Bot: 'Book what?'
• With Memory:
• • Bot remembers recent context
• • Can handle follow-up questions
• • User: 'How much is a haircut?' -> Bot: '$30'
• • User: 'Book one for tomorrow' -> Bot: 'I'll book a haircut for tomorrow. What time works?'

Configuring Memory in n8n

• Two Types of Memory:
• 1. Window Memory (Short-term):
• • Remembers last X messages
• • Good for simple conversations
• • Example: Last 10 messages
• 2. Buffer Memory (Session-based):
• • Remembers entire conversation session
• • Clears when conversation ends
• • Good for support chats
• Adding Memory in n8n:
• 1. Click AI Agent node
• 2. Look for 'Memory' section
• 3. Add 'Window Buffer Memory'
• 4. Set window size (usually 10-20 messages)
• Now your bot can have real conversations!