Amazon Bedrock - Creating a AI Agent

Creating an AI agent using Amazon Bedrock involves configuring foundation models, defining tools, and deploying an agent capable of interacting with users and systems. Below is a step-by-step guide with visual references and a clear explanation.

Step 1: Understand Amazon Bedrock ArchitectureCore components:

1. Foundation Model (FM) – Claude, Titan, Llama, etc.

2. Agent – Orchestrates reasoning and actions

3. Knowledge Base – Connects to your data (S3, PDFs, databases)

4. Action Groups – Lambda functions for performing tasks

5. Client Application – Web app, chatbot, or API

Step 2: Enable Amazon Bedrock

To enable the Amazon Bedrock - 

1. Login to AWS Console

2. Search → Amazon Bedrock

3. Click → Model access
   

4. Enable models like:

• Claude (best for agents)

• Titan

• Llama

Step 3: Create Knowledge Base (Optional but Recommended)

A knowledge base allows agents to use your custom data. To create a Knowledge Base - 

Steps:

1. Go to Bedrock Console

2. Click → Knowledge Bases

3. Click → Create knowledge base

4. Select:

   • Data source → S3 bucket

   • Embedding model → Titan embeddings
     

5. Sync data

Step 4: Create Lambda Function (Action)

Creating a basic lambda to say hello 😊

Example Lambda (Python):

def lambda_handler(event, context):
    user_input = event['inputText']

    if "hello" in user_input.lower():
        return {
            "message": "Hello! How can I help you?"
        }

    return {
        "message": "You said: {user_input}"
    }

Step 5: Create an Agent in Bedrock

1. Go to Bedrock Console

2. Click → Agents

3. Click → Create Agent

And fill in the details as below - 

Agent Name: MySupportAgent

Foundation Model: Claude 3 Sonnet or Claude 3 Haiku

Instructions (Important): "You are a helpful AI assistant that helps users answer questions and perform actions using provided tools."

Step 6: Add Action Group

Follow the steps below to add an Action Group:

1. Open your agent

2. Click → Add Action Group

3. Select:

   • Lambda function created earlier

4. Define schema

Example schema:

{
  "openapi": "3.0.0",
  "paths": {
    "/hello": {
      "post": {
        "operationId": "sayHello",
        "requestBody": {
          "content": {
            "application/json": {}
          }
        },
        "responses": {
          "200": {
            "description": "Success"
          }
        }
      }
    }
  }
}

Step 7: Test the Agent

Now the final step. It's time to test your agent. Follow the steps below - 

Steps:

1. Click → Test Agent

2. Enter message:

Hello agent

The agent will respond using model + Lambda.

Your agent must respond with a message such as "Hello! How can I help you?"

𝐼𝒻 𝓎𝑜𝓊 𝓌𝑜𝓊𝓁𝒹 𝓁𝒾𝓀𝑒 𝓉𝑜 𝒸𝑜𝓃𝓉𝓇𝒾𝒷𝓊𝓉𝑒, 𝓅𝓁𝓏 𝓌𝓇𝒾𝓉𝑒 & E𝓂𝒶𝒾𝓁 𝓎𝑜𝓊𝓇 𝒶𝓇𝓉𝒾𝒸𝓁𝑒 𝓉𝑜 𝒹𝓇𝑜𝒾𝒹𝒶𝓂𝒶𝓇𝟢𝟢𝟩@𝑔𝓂𝒶𝒾𝓁.𝒸𝑜𝓂.  𝒴𝑜𝓊 𝒸𝒶𝓃 𝓈𝑒𝑒 𝓎𝑜𝓊𝓇 𝒶𝓇𝓉𝒾𝒸𝓁𝑒 𝒶𝓅𝓅𝑒𝒶𝓇𝒾𝓃𝑔 𝑜𝓃 𝓉𝒽𝑒 𝓂𝒶𝒾𝓃 𝓅𝒶𝑔𝑒 𝒶𝓃𝒹 𝒽𝑒𝓁𝓅𝒾𝓃𝑔 𝑜𝓉𝒽𝑒𝓇𝓈. 

꧁༺ 𝐻𝑜𝓅𝑒 𝓉𝒽𝒾𝓈 𝓌𝒾𝓁𝓁 𝒽𝑒𝓁𝓅 𝓎𝑜𝓊. 𝒦𝑒𝑒𝓅 𝓋𝒾𝓈𝒾𝓉𝒾𝓃𝑔 𝓂𝓎 𝒷𝓁𝑜𝑔 :) ༻꧂