Amazon Bedrock
By Amazon Web Services
Amazon Bedrock is a fully managed AWS service that gives developers unified API access to foundation models from multiple AI providers, letting them build generative AI applications without hosting or fine-tuning models themselves.
Definition
Amazon Bedrock is a fully managed AWS service that gives developers unified API access to foundation models from multiple AI providers, letting them build generative AI applications without hosting or fine-tuning models themselves.
Overview
Before services like Bedrock existed, using a foundation model in an application usually meant integrating with a single provider's own API and infrastructure. Bedrock instead offers one managed service where AWS customers can call foundation models from several providers — including Anthropic's Claude family, Meta's Llama, Mistral AI models, and Amazon's own Titan/Nova models — through a consistent API, while the underlying compute, scaling, and availability are managed by AWS. Beyond raw model access, Bedrock provides capabilities layered on top of the models themselves: fine-tuning and customization on private data, knowledge-base integration for retrieval-augmented generation, guardrails for filtering unwanted content, and an agents framework for building multi-step, tool-using AI workflows. Because it runs inside a customer's AWS environment, Bedrock is often chosen by organizations that want generative AI capability without sending data outside their existing AWS security and compliance boundary, and it's frequently paired with Amazon SageMaker for teams that also need custom model training. The concepts behind foundation models and retrieval-augmented generation are covered in the Large Language Models course.
Key Features
- Unified API access to foundation models from multiple AI providers
- Fully managed infrastructure — no need to host or scale models yourself
- Fine-tuning and customization of models on private data
- Built-in knowledge bases for retrieval-augmented generation (RAG)
- Guardrails for content filtering and responsible AI controls
- Agents framework for building multi-step, tool-using AI applications
- Runs inside a customer's AWS account and security boundary