Create Your First Agent

Agents are most useful when they are specific. This tutorial shows how to define that specificity in instructions, data access, and testing rather than relying on a generic assistant prompt.

What Are Agents?

Agents are specialized AI assistants that you can customize for specific tasks, roles, or domains. They can:

• Have custom personalities and communication styles
• Access specific sets of documents and knowledge
• Follow particular instructions or workflows
• Be shared with specific users or groups

Think of agents as different “experts” you can consult - like having a financial analyst, technical writer, or customer service representative always available.

Step 1: Access Agent Management

1. Navigate to the Agents section in Simple Chat
2. Click “Create New Agent” or the “+” button
3. You’ll see the agent creation interface

Step 2: Configure Basic Agent Settings

Agent Identity

1. Name your agent: Choose a clear, descriptive name
   • Good examples: “Financial Analyst”, “Technical Documentation Helper”, “Customer Support Bot”
   • Avoid generic names like “Agent 1”
2. Add a description: Explain what this agent does
   • Example: “Analyzes financial reports and provides insights on budget performance and trends”
3. Choose an avatar/icon (if available): Visual representation helps users understand the agent’s purpose

Agent Instructions (System Prompt)

This is where you define your agent’s personality and capabilities:

You are a Financial Analyst Agent specializing in budget analysis and financial reporting. 

Your role:
- Analyze financial documents and spreadsheets
- Provide insights on budget performance, trends, and variances
- Explain financial concepts in clear, business-friendly language
- Recommend actionable next steps based on data

Your personality:
- Professional but approachable
- Detail-oriented and analytical
- Ask clarifying questions when needed
- Always provide specific examples and evidence

When responding:
1. Start with a brief summary of your findings
2. Provide detailed analysis with specific numbers
3. Highlight key concerns or opportunities
4. Suggest concrete next steps

Step 3: Configure Agent Knowledge

Workspace Access

1. Select workspaces the agent can access:
   • Your personal workspace
   • Specific group workspaces
   • Public workspaces (if available)
2. Choose relevant documents:
   • Financial reports for a Financial Analyst agent
   • Technical docs for a Documentation agent
   • Product catalogs for a Sales agent

Knowledge Scope

Consider what information your agent needs:

• Broad access: Good for general-purpose agents
• Focused access: Better for specialized agents to avoid confusion

Step 4: Test Your Agent

Initial Testing

1. Save your agent configuration
2. Start a new conversation with your agent
3. Test basic functionality:
   • Ask about the agent’s role: “What can you help me with?”
   • Test knowledge access: “What documents do you have access to?”
   • Try a representative task

Example Test Conversation

You: “Hello! What can you help me with?”

Agent: “Hi! I’m your Financial Analyst Agent. I specialize in analyzing financial documents and providing insights on budget performance and trends. I can help you with:

• Budget variance analysis
• Financial report interpretation
• Trend identification and forecasting
• ROI calculations and recommendations

What financial questions can I help you explore today?”

You: “Can you analyze our Q3 budget performance?”

Agent: [Searches relevant financial documents and provides analysis]

Step 5: Refine and Improve

Based on your testing, you might need to adjust:

Common Adjustments

• Instructions too broad: Make them more specific
• Agent too formal/informal: Adjust personality in system prompt
• Wrong knowledge access: Modify workspace permissions
• Responses off-topic: Refine the role definition

Iterative Improvement

1. Test with real use cases
2. Note what works and what doesn’t
3. Update instructions and configuration
4. Test again

Step 6: Share Your Agent (Optional)

If your agent works well:

1. Configure sharing settings
2. Add team members who can access the agent
3. Provide usage guidelines for your team

Real-World Agent Examples

Technical Documentation Agent

You are a Technical Documentation Specialist focused on helping developers and engineers find and understand technical information.

Your expertise:
- API documentation and integration guides
- Troubleshooting technical issues
- Code examples and best practices
- Architecture and system design

Always provide:
- Step-by-step instructions
- Code examples when relevant
- Links to relevant documentation sections
- Warnings about common pitfalls

Customer Service Agent

You are a Customer Service Agent dedicated to helping users with product questions and support issues.

Your approach:
- Friendly, patient, and helpful tone
- Listen carefully to customer concerns
- Provide clear, actionable solutions
- Know when to escalate to human agents

Your knowledge includes:
- Product features and capabilities
- Common troubleshooting steps
- Company policies and procedures
- FAQ and knowledge base articles

Best Practices for Agent Creation

Do’s

• ✅ Give agents specific, clear roles
• ✅ Test thoroughly before sharing
• ✅ Update instructions based on feedback
• ✅ Provide relevant document access
• ✅ Use consistent personality and tone

Don’ts

• ❌ Make instructions too vague or generic
• ❌ Give access to irrelevant documents
• ❌ Create agents that duplicate existing functionality
• ❌ Forget to test edge cases
• ❌ Leave agents without clear purpose

What You’ve Accomplished

Congratulations! You’ve successfully: ✅ Created a custom agent with specific expertise
✅ Configured agent personality and behavior
✅ Set up appropriate knowledge access
✅ Tested and refined your agent
✅ Learned agent best practices

Next Steps

Now that you have your first agent:

• Upload and organize more documents to expand agent knowledge
• Classify documents to help agents find relevant information
• Create additional agents for different use cases
• Explore advanced agent configuration options

Troubleshooting

Agent not finding relevant information?

• Check workspace access permissions
• Verify documents are properly uploaded and processed
• Make sure agent instructions are specific enough

Agent responses too generic?

• Add more specific instructions about tone and approach
• Include examples of desired responses
• Narrow the agent’s knowledge scope

Agent not staying in character?

• Strengthen the personality section in instructions
• Add reminders about the agent’s role
• Test with edge cases to identify instruction gaps

Next: Uploading Documents →