Published on

Amazon unveils Bedrock, a cloud-based platform for AI-generated content including text and images


Amazon is entering the generative AI arena with Bedrock, a new API for Amazon Web Services (AWS) that enables developers to access and tailor AI tools for generating text or images. Bedrock serves as a cloud-based, configurable alternative to OpenAI's ChatGPT and DALL-E 2, targeting businesses and developers.

With Bedrock, AWS customers can create written content, develop chatbots, summarize text, classify images, and more using text prompts. Users can choose from Amazon's Titan foundation model (FM) and various startup models, such as Anthropic's Claude (a Google-backed ChatGPT competitor developed by former OpenAI employees), AI21's Jurassic-2 (a language model specializing in Spanish, French, German, Portuguese, Italian, and Dutch), and Stable Diffusion (a widely-used open-source image generator). Businesses and developers can also customize the models' behavior based on input, which Amazon states will not be utilized for training the models, addressing a key privacy concern for companies entering sensitive data.

Amazon sees the variety of AI models available as a means to provide flexibility to its customers. The company's description emphasizes that with Bedrock's serverless experience, users can quickly start, privately customize FMs with their data, and seamlessly integrate and deploy them into their applications using familiar AWS tools and features (including Amazon SageMaker ML features like Experiments for testing different models and Pipelines for managing FMs at scale) without managing any infrastructure.

Amazon CEO Andy Jassy explained to CNBC that most companies want to use large language models but are deterred by the high costs and lengthy timeframes involved in training them. Bedrock offers a solution by allowing businesses to build upon an already robust foundational model and customize it to suit their needs.