Introduction to AWS Bedrock

Duration: 45 min

AWS Bedrock is a fully managed service that provides access to foundation models (FMs) via a simple API. Instead of managing infrastructure or fine-tuning models yourself, you invoke pre-trained models from leading AI companies like Anthropic, Meta, Mistral, and Stability AI. This module covers what Bedrock is, the models available, pricing, and real-world use cases.

What is AWS Bedrock?

Bedrock abstracts away the complexity of deploying and managing foundation models. You get:

• No infrastructure management — AWS handles scaling, availability, and updates
• Multiple model providers — Claude, Llama, Mistral, Titan, Stable Diffusion in one place
• Simple API — Invoke models with a few lines of Python
• Security & compliance — Data stays in your AWS account; no model training on your data by default

Foundation Models Explained

A foundation model is a large neural network trained on massive amounts of text, code, or images. They're "foundational" because they can be adapted to many downstream tasks without retraining from scratch.

Key characteristics:

• Trained on billions of parameters
• Learned from diverse, unlabeled data
• Can perform multiple tasks (text generation, summarization, Q&A, code generation)
• Improved with prompt engineering and fine-tuning

Bedrock Pricing Model

Bedrock uses on-demand pricing (pay per token) or provisioned throughput (reserved capacity):

{
  "on_demand": {
    "claude_3_sonnet": {
      "input_tokens": "$0.003 per 1K tokens",
      "output_tokens": "$0.015 per 1K tokens"
    },
    "llama_2_70b": {
      "input_tokens": "$0.00195 per 1K tokens",
      "output_tokens": "$0.00256 per 1K tokens"
    }
  },
  "provisioned_throughput": {
    "description": "Reserve capacity for predictable workloads",
    "commitment": "1 or 6 month terms",
    "discount": "Up to 40% savings vs on-demand"
  }
}

Common Use Cases

1. Customer Support Chatbots

• Automated Q&A with company knowledge bases
• Reduced support ticket volume
• 24/7 availability

2. Content Generation

• Blog posts, product descriptions, marketing copy
• Personalized email campaigns
• Code generation and documentation

3. Data Analysis & Insights

• Summarize documents and reports
• Extract key information from unstructured data
• Generate business intelligence

4. Code Assistance

• Code review and refactoring suggestions
• Bug detection and fixes
• Documentation generation

5. Retrieval-Augmented Generation (RAG)

• Answer questions using private data
• Reduce hallucinations with grounded context
• Build domain-specific AI assistants

Getting Started with Bedrock

Prerequisites

# Install AWS CLI v2
curl "https://awscli.amazonaws.com/awscli-exe-macos.zip" -o "awscliv2.zip"
unzip awscliv2.zip
sudo ./aws/install

# Install boto3
pip install boto3

# Configure AWS credentials
aws configure

Enable Model Access

Models in Bedrock are disabled by default. Enable them in the AWS Console:

# List available models
aws bedrock list-foundation-models --region us-east-1

# Or enable via console: Bedrock > Model access > Request access

First API Call

import boto3

client = boto3.client('bedrock-runtime', region_name='us-east-1')

response = client.invoke_model(
    modelId='anthropic.claude-3-sonnet-20240229-v1:0',
    body=json.dumps({
        "anthropic_version": "bedrock-2023-06-01",
        "max_tokens": 1024,
        "messages": [
            {
                "role": "user",
                "content": "What is AWS Bedrock?"
            }
        ]
    })
)

result = json.loads(response['body'].read())
print(result['content'][0]['text'])

Bedrock vs Alternatives