Creating Your First Agent

10-minute tutorial

Follow this step-by-step guide to create and deploy your first AI agent using Motte.

Prerequisites

Before you start:

✅ API keys configured (see API Keys Setup)
✅ Basic understanding of AI agents
✅ 10 minutes of your time

Step 1: Choose Your Agent Type

For your first agent, we'll create a Research Assistant that can gather information and provide summaries.

Research Agent

Gathers and analyzes information from various sources

Recommended

Writing Agent

Creates content based on prompts and guidelines

Analysis Agent

Processes data and provides insights

Step 2: Set Up Memory (MotteMB)

First, let's give our agent some memory so it can learn and remember context.

Navigate to MotteMB

1. Go to Tools → MotteMB
2. Click "Create Memory"
3. Add initial context:

"I am a research assistant specialized in gathering 
information on technology topics. I provide accurate, 
well-sourced summaries and can identify key trends 
and insights from multiple sources."

Pro tip: The more context you provide in the initial memory, the better your agent will perform. Include its role, capabilities, and any specific instructions.

Step 3: Create a Workflow (MotteAF)

Now let's create a simple workflow for our research agent.

Workflow Steps:

Navigate to MotteAF
Click "New Flow"
Name it "Research Assistant Workflow"
Drag a Research Agent to the canvas
Connect it to an Analysis Agent
Add a Writer Agent for the final summary

Visual Workflow:

┌─────────────┐    ┌─────────────┐    ┌─────────────┐
│   Research  │───▶│   Analysis  │───▶│   Writer    │
│    Agent    │    │    Agent    │    │    Agent    │
└─────────────┘    └─────────────┘    └─────────────┘
      │                    │                    │
      ▼                    ▼                    ▼
   Gather Info        Process Data        Create Summary

Step 4: Configure Tools (MotteTF)

Let's connect some external tools to enhance our agent's capabilities.

Web Search Tool

Endpoint: https://api.search.com/v1

Purpose: Gather current information

Authentication: API Key

Wikipedia API

Endpoint: https://en.wikipedia.org/api

Purpose: Get detailed background info

Authentication: None required

Step 5: Test Your Agent

Ready to test!

Your agent is now configured. Let's run a test to see it in action.

Test Prompt:

"Research the latest developments in artificial intelligence and provide a summary of the top 3 trends in 2024."

Testing Steps:

Go back to MotteAF
Select your "Research Assistant Workflow"
Click "Run Flow"
Enter the test prompt above
Watch the agents work in sequence
Review the final output

Step 6: Monitor Performance (MotteAW)

While your agent is running, you can monitor its performance in real-time.

What to watch for:

Response times: How quickly each agent responds
Memory usage: How the agent accesses stored context
Tool calls: Which external APIs are being used
Error rates: Any failures or retries
Output quality: Relevance and accuracy of responses

Expected Results

Your agent should:

✅ Search for current AI developments
✅ Analyze and identify top trends
✅ Provide a well-structured summary
✅ Include relevant sources and data
✅ Complete the task in under 2 minutes

Troubleshooting

Agent not responding?

Check your API keys and ensure you have sufficient credits.

Poor quality responses?

Add more specific context to your agent's memory and refine your prompts.

Workflow not executing?

Ensure all agents are properly connected and your workflow has a clear start point.

🎉 Congratulations!

You've successfully created and deployed your first AI agent using Motte. Your research assistant can now gather information, analyze it, and provide structured summaries.