A late but expected guest to the party, Amazon has introduced Amazon Bedrock, a platform for generative AI-powered apps using pre-trained models from third-party startups.
Bedrock offers access to AWS in-house trained Titan FMs (foundation models) and is currently available in “limited preview.”
Customers can access AI models from various providers, including AWS, via an API.
Third-party models on Bedrock include AI21 Labs’ Jurassic-2 family, Anthropic’s Claude, and Stability AI’s suite of text-to-image models.
The Titan FM family currently consists of a text-generating model and an embedding model.
- Text-generating model: This model is designed to generate human-like text based on a given input. It can perform tasks such as writing blog posts, emails, summarising documents, and extracting information from databases. The text-generating model is similar in concept to OpenAI’s GPT-4 but may not necessarily match its performance.
- Embedding model: This model is responsible for converting text inputs, like words and phrases, into numerical representations called embeddings. These embeddings capture the semantic meaning of the text, making it easier for the AI to understand and process the input.
AWS customers can customize Bedrock models with as few as 20 labelled examples stored in Amazon S3.
Amazon S3 (Simple Storage Service) is a scalable, object storage service provided by Amazon Web Services (AWS), widely used by businesses and developers for various purposes, including backup and recovery, big data analytics, content storage and distribution, data archiving, and more.
Amazon has also made CodeWhisperer, its AI-powered code-generating service, free for developers without usage restrictions.
CodeWhisperer is designed to assist with tasks like autocompleting code, suggesting code snippets, and providing context-aware recommendations. It was introduced as a response to GitHub’s Copilot.