AGENTIC AI TOOL USE

Build AI Agents That Think and Act

Create intelligent autonomous assistants that can reason, use tools, and accomplish complex tasks

AI agents go beyond simple chat interactions. They can analyze complex scenarios, break down multi-step tasks, use external tools, and work autonomously to achieve goals - all while maintaining context and learning from your organization's knowledge.

2
Execution modes
10+
LLM models
Built-in
Tools included
Real-time
Streaming

Video Tutorial

What Makes Agents Different

Understanding the key differences between traditional chat and agentic AI helps you choose the right approach for your use case.

Capability Chat Interface AI Agent
Response Type Single response ✓ Multi-step reasoning
Tool Usage Limited or none ✓ Full tool ecosystem
Task Planning Basic understanding ✓ Complex task breakdown
Memory Conversation only ✓ Knowledge base access
Actions Text responses ✓ External system integration
Autonomy Reactive ✓ Proactive problem solving

Execution Modes

Choose the right execution mode based on task complexity and response time requirements.

Standard Mode

Fast & Efficient

Optimized for quick responses and straightforward tasks. Best for customer support, basic queries, and simple tool usage.

  • Sub-second response times
  • Lower computational cost
  • Single-step reasoning
  • Perfect for high-volume use

Super Mode

Deep Reasoning

Advanced cognitive capabilities for complex problem-solving, multi-step analysis, and sophisticated task planning.

  • Multi-step reasoning chains
  • Complex task decomposition
  • Advanced tool orchestration
  • Self-correction capabilities

Creating an Agent

Follow these steps to build your first AI agent.

  1. Define Agent Purpose Clearly outline what tasks your agent will handle and its primary objectives.
  2. Choose Execution Mode Select Standard for quick responses or Super for complex reasoning tasks.
  3. Select LLM Model Pick from Claude 3.5 Sonnet, GPT-4, or other available models based on your needs.
  4. Configure System Prompt Write detailed instructions that define the agent's personality, behavior, and expertise.
  5. Add Knowledge Base Connect relevant documents and data sources for context-aware responses.
  6. Enable Tools Grant access to web search, file operations, API integrations, and custom MCP servers.
  7. Set Temperature Adjust creativity level from 0 (precise) to 1 (creative) based on use case.
  8. Test & Deploy Validate agent responses and deploy for use in workflows, widgets, or direct access.

System Prompt Templates

Well-crafted system prompts are crucial for agent performance. Here's a customer support agent example:

You are a helpful customer support agent for TechCorp, a software company. Your role: - Provide accurate, friendly assistance to customers - Use the knowledge base to find relevant information - Create support tickets for complex issues - Always maintain a professional, empathetic tone Guidelines: - Greet customers warmly and ask how you can help - Search the knowledge base before answering questions - If you cannot find information, create a support ticket - Provide step-by-step instructions when helping with technical issues - Always ask if there's anything else you can help with Available tools: - search_knowledge_base: Search company documentation - create_support_ticket: Create tickets for unresolved issues - send_email: Send follow-up emails to customers Remember: Your goal is to resolve customer issues quickly and effectively while maintaining a positive experience.
Pro Tip

Include specific examples of how to handle edge cases and define clear boundaries for what the agent should and shouldn't do.

Agent Templates

Start with these proven agent configurations for common business scenarios.

Support Agent

24/7 customer support with knowledge base integration, ticket creation, and escalation workflows.

Standard Mode Temperature: 0.3

Data Analyst

Query databases, generate reports, and provide insights from your business data with natural language.

Super Mode Temperature: 0.1

Research Agent

Comprehensive research with web search, document analysis, and structured report generation.

Super Mode Temperature: 0.5

Web Automation

Automate web interactions, form filling, and data extraction with browser control tools.

Standard Mode Temperature: 0.2

Model Selection Guide

Choose the right LLM model based on your specific requirements and budget.

Model Best For Cost Speed Reasoning
Claude 3.5 Sonnet General purpose, coding, analysis Medium Fast Excellent
Claude 3 Haiku High-volume, quick responses Low Very Fast Good
GPT-4 Turbo Complex reasoning, creativity High Medium Excellent
GPT-3.5 Turbo Cost-effective, simple tasks Low Fast Good
Local Models Privacy, offline, custom Variable Variable Variable

Temperature Guide

Temperature controls the randomness and creativity of AI responses. Choose the right setting for your use case.

Low (0.0 - 0.3)

Precise & Consistent

Use for factual responses, data analysis, customer support, and tasks requiring accuracy and consistency.

Medium (0.4 - 0.7)

Balanced

Ideal for general conversation, content creation, and tasks that benefit from some creativity while maintaining coherence.

High (0.8 - 1.0)

Creative & Varied

Best for creative writing, brainstorming, marketing content, and scenarios where diverse responses are valuable.

Important

Higher temperatures increase response variability but may reduce factual accuracy. Always test your agent's performance at different temperature settings.

Troubleshooting

Common issues and solutions when working with AI agents.

Agent responses are inconsistent

Cause: Temperature too high or vague system prompt

Solution: Lower temperature to 0.1-0.3 and provide more specific instructions in system prompt

Agent doesn't use available tools

Cause: Tools not mentioned in system prompt or unclear tool descriptions

Solution: Explicitly list available tools in system prompt and provide clear usage examples

Responses are too slow

Cause: Super Mode enabled for simple tasks or complex knowledge base

Solution: Switch to Standard Mode for simple tasks or optimize knowledge base size

Agent ignores knowledge base content

Cause: Knowledge base not properly connected or poor document quality

Solution: Verify knowledge base connection and ensure documents are well-formatted with clear structure

Ready to Build Your First Agent?

Create intelligent AI agents that can handle complex tasks and integrate with your business systems.