Amazon Bedrock FAQs

Amazon Bedrock customers can choose from some of the most cutting-edge FMs available today. This includes models from:

• AI21 Labs

• Amazon

• Anthropic

• Cohere

• DeepSeek

• Luma AI

• Meta

• Mistral AI

• poolside (coming soon)

• Stability AI

• TwelveLabs (coming soon)

See here for supported foundation models from each provider:
https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html

There are five reasons to use Amazon Bedrock for building generative AI applications.

• Choice of leading FMs: Amazon Bedrock offers an easy-to-use developer experience to work with a broad range of high-performing FMs from Amazon and leading AI companies like AI21 Labs, Anthropic, Cohere, Meta, Mistral AI, and Stability AI. You can quickly experiment with a variety of FMs in the playground, and use a single API for inference regardless of the models you choose, giving you the flexibility to use FMs from different providers and keep up to date with the latest model versions with minimal code changes.

• Easy model customization with your data: Privately customize FMs with your own data through a visual interface without writing any code. Simply select the training and validation data sets stored in Amazon Simple Storage Service (Amazon S3) and, if required, adjust the hyperparameters to achieve the best possible model performance.

• Fully managed agents that can invoke APIs dynamically to execute tasks: Build agents that execute complex business tasks—from booking travel and processing insurance claims to creating ad campaigns, preparing tax filings, and managing your inventory—by dynamically calling your company systems and APIs. Fully managed agents for Amazon Bedrock extend the reasoning capabilities of FMs to break down tasks, create an orchestration plan, and execute it.

• Native support for RAG to extend the power of FMs with proprietary data: With Amazon Bedrock Knowledge Bases, you can securely connect FMs to your data sources for retrieval augmentation—from within the managed service—extending the FM’s already powerful capabilities and making it more knowledgeable about your specific domain and organization.

• Data security and compliance certifications: Amazon Bedrock offers several capabilities to support security and privacy requirements. Amazon Bedrock is in scope for common compliance standards such as Service and Organization Control (SOC), International Organization for Standardization (ISO), is Health Insurance Portability and Accountability Act (HIPAA) eligible, and customers can use Amazon Bedrock in compliance with the General Data Protection Regulation (GDPR). Amazon Bedrock is CSA Security Trust Assurance and Risk (STAR) Level 2 certified, which validates the use of best practices and the security posture of AWS cloud offerings. With Amazon Bedrock, your content is not used to improve the base models and is not shared with any model providers. Your data in Amazon Bedrock is always encrypted in transit and at rest, and you can optionally encrypt the data using your own keys. You can use AWS PrivateLink with Amazon Bedrock to establish private connectivity between your FMs and your Amazon Virtual Private Cloud (Amazon VPC) without exposing your traffic to the Internet.

You can quickly get started with use cases:

• Create new pieces of original content, such as short stories, essays, social media posts, and web page copy.

• Search, find, and synthesize information to answer questions from a large corpus of data.

• Create realistic and artistic images of various subjects, environments, and scenes from language prompts.

• Help customers find what they’re looking for with more relevant and contextual product recommendations than word matching.

• Get a summary of textual content such as articles, blog posts, books, and documents to get the gist without having to read the full content.

• Suggest products that match shopper preferences and past purchases

Explore more generative AI use cases.

Available in public preview, latency-optimized inference in Amazon Bedrock offers reduced latency without compromising accuracy. As verified by Anthropic, with latency-optimized inference on Amazon Bedrock, Claude 3.5 Haiku runs faster on AWS than anywhere else. Additionally, with latency-optimized inference in Bedrock, Llama 3.1 70B and 405B runs faster on AWS than any other major cloud provider. Using purpose-built AI chips like AWS Trainium2 and advanced software optimizations in Amazon Bedrock, customers can access more options to optimize their inference for a particular use case.

Key Features:

• Reduces response times for foundation model interactions

• Maintains accuracy while improving speed

• Requires no additional setup or model fine-tuning

Supported Models: Anthropic's Claude 3.5 Haiku and Meta's Llama 3.1 models 405B and 70B

Availability: The US East (Ohio) Region via cross-region inference

To get started, visit the Amazon Bedrock console. For more information visit the Amazon Bedrock documentation.

Accessing the latency-optimized inference in Amazon Bedrock requires no additional setup or model fine-tuning, allowing for immediate enhancement of existing generative AI applications with faster response times. You can toggle on the “Latency optimized” parameter while invoking the Bedrock inference API.

To get started, visit the Amazon Bedrock console. For more information visit the Amazon Bedrock documentation.

With agents, developers have seamless support for monitoring, encryption, user permissions, versioning, and API invocation management without writing custom code. Amazon Bedrock Agents automate the prompt engineering and orchestration of user-requested tasks. Developers can use the agent-created prompt template as a baseline to further refine it for an enhanced user experience. They can update the user input, orchestration plan, and the FM response. With access to the prompt template developers have better control over the Agent orchestration.

With fully managed agents, you don’t have to worry about provisioning or managing infrastructure and can take applications to production faster.

Guardrails help you to define a set of six policies to help safeguard your generative AI applications. You can configure the following policies in Amazon Bedrock Guardrails:

• Multi modal content filters – Configure thresholds to detect and filter harmful text and/or image content across categories including hate, insults, sexual, violence, misconduct, and prompt attacks.

• Denied topics – Define a set of topics that are undesirable in the context of your application. The filter will help block them if detected in user queries or model responses.

• Word filters – Configure filters to help block undesirable words, phrases, and profanity (exact match). Such words can include offensive terms, competitor names, etc.

• Sensitive information filters – Configure filters to help block or mask sensitive information, such as personally identifiable information (PII), or custom regex in user inputs and model responses. Blocking or masking is done based on probabilistic detection of sensitive information in standard formats in entities such as SSN number, Date of Birth, address, etc. This also allows configuring regular expression based detection of patterns for identifiers.

• Contextual grounding check – Help detect and filter hallucinations if the responses are not grounded (e.g., factually inaccurate or new information) in the source information and irrelevant to user’s query or instruction.

• Automated Reasoning checks – Help detect factual inaccuracies in generated content, suggest corrections, and explain why responses are accurate by checking against a structured, mathematical representation of knowledge called an Automated Reasoning Policy.

There are five guardrail policies each with different off-the-shelf protections

• Content filters – This has 6 off the shelf categories (hate, insults, sexual, violence, misconduct (incl. criminal activity) and prompt attack (jailbreak and prompt injection. Each category can have further customized thresholds in terms of aggressiveness of filtering - low/medium/high for both text and image content.

• Denied topic – These are customized topics that customers can define using simple natural language description

• Sensitive information filter – These come with 30+ off the shelf PIIs. It can be further customized by adding customer’s proprietary information that are sensitive.

• Word filters – It comes with off the shelf profanity filtering and can be further customized with custom words.

• Contextual grounding checks – It can help detect hallucinations for RAG, summarization, and conversational applications, where source information can be used as reference to validate the model response.

Documents

Bedrock Data Automation supports both standard output and custom output for documents.

• Standard output will provide extraction of text from documents and generative output such as document summary and captions for tables/figures/diagrams. Output is returned in reading order and can optionally be grouped by layout element, which will include headers/footers/titles/tables/figures/diagrams. Standard output will be used for BDA integration with Bedrock Knowledge Bases.

• Custom Output leverages blueprints, which specify output requirements using natural language or a schema editor. Blueprints include a list of fields to extract and a data format for each field.

Bedrock Data Automation supports PDF, PNG, JPG, TIFF, a max of 1500 pages, and a max file size of 500MB per API request. By default, BDA will support 50 concurrent jobs and 10 transactions per second per customer.

Images

Bedrock Data Automation supports both standard output and custom output for images.

• Standard output will provide summarization, detected explicit content, detected text, logo detection and Ad taxonomy: IAB for images. Standard output will be used for BDA integration with Bedrock Knowledge Bases.

• Custom Output leverages blueprints, which specify output requirements using natural language or a schema editor. Blueprints include a list of fields to extract and a data format for each field.

Bedrock Data Automation supports JPG, PNG, a max resolution of 4K, and a max file size of 5 MB per API request. By default, BDA supports a max concurrency of 20 images at 10 transactions per second (TPS) per customer.

Videos

Bedrock Data Automation supports both standard output for videos.

• Standard output will provide full video summary, chapter segmentation, chapter summary, full audio transcription, speaker identification, detected explicit content, detected text, logo detection and Interactive Advertising Bureau (IAB) taxonomy for videos. Full video summary is optimized for content with descriptive dialogue such as product overviews, trainings, news casts, and documentaries.

Bedrock Data Automation supports MOV and MKV with H.264, VP8, VP9, a max video duration of 4 hours, and a max file size of 2 GB per API request. By default, BDA supports a max concurrency of 20 videos at 10 transactions per second (TPS) per customer.

Audio

Bedrock Data Automation supports both standard output for audio.

• Standard output will provide summarization including chapter summarization, full transcription, and detect explicit content moderation for audio files.

Bedrock Data Automation supports FLAC, M4A, MP3, MP4, Ogg, WebM, WAV, a max audio duration of 4 hours, and a max file size of 2 GB per API request.

To access Amazon Bedrock's capabilities within Amazon SageMaker Unified Studio, developers and their admins will need to follow these steps:

1. Create a new domain in Amazon SageMaker Unified Studio.

2. Enable the Gen AI application development project profile.

3. Access Amazon Bedrock through the Generative AI Playground (Discover) and Generative AI App Development (Build) sections, using their company's single sign-on (SSO) credentials within Amazon SageMaker Unified Studio.

Amazon Bedrock's capabilities in Amazon SageMaker Unified Studio are ideal for enterprise teams who need a governed environment for collaboratively building and deploying generative AI applications. Through Amazon SageMaker Unified Studio, teams can access:
 

• The Generative AI Playground in the Discover section enables teams to experiment with foundation models (FMs), test different models and configurations, compare model outputs, and collaborate on prompts and applications. This environment provides a seamless way for teams to evaluate and understand the capabilities of different models before implementing them in their applications.
   

• The Generative AI App Development in the Build section provides teams with the tools needed to create production-ready generative AI applications. Teams can create and manage Knowledge Bases, implement Guardrails for responsible AI, develop Agents and Flows, and collaborate securely while maintaining governance and compliance controls. This environment is particularly valuable for organizations that require secure collaboration and seamless access to Amazon Bedrock's full range of capabilities while maintaining enterprise security and compliance standards.