Answer: Built-in model capabilities such as summarization

## Explanation **Correct Answer: B - Built-in model capabilities such as summarization** Amazon Bedrock is a fully managed service that offers access to foundation models (FMs) from leading AI companies through a single API. The key advantage of Bedrock is that it provides **pre-built capabilities** like summarization, classification, question answering, and more, without requiring customers to design or train their own models from scratch. **Why Option B is correct:** - **Built-in capabilities**: Amazon Bedrock foundation models come with inherent capabilities for common tasks like summarization, translation, content generation, etc. - **No custom model design needed**: The company can simply use the summarization capability of available foundation models without building their own model. - **Legal document summarization**: Foundation models in Bedrock can be prompted to summarize legal documents effectively. **Why Option A is incorrect:** - **Bedrock Embeddings API**: This is used for creating vector embeddings (numerical representations) of text for semantic search and retrieval, not for summarization tasks. - **Different purpose**: Embeddings are primarily used for similarity search, clustering, and recommendation systems, not for generating summarized content. **Key AWS Bedrock Concepts:** 1. **Foundation Models**: Pre-trained large language models that can perform various tasks 2. **Built-in capabilities**: Ready-to-use functionalities like summarization, classification, etc. 3. **Prompt engineering**: The company would need to provide appropriate prompts to get the desired summarization output 4. **Managed service**: No infrastructure management required - Bedrock handles scaling, availability, and maintenance **Use Case Fit:** For summarizing legal documents, the company would: 1. Choose an appropriate foundation model in Bedrock 2. Use the model's built-in summarization capability 3. Provide the legal document text as input with appropriate prompts 4. Receive the summarized output without any model training or custom development

Author: Ritesh Yadav