AWS offers new generative AI options, makes Bedrock available

Spread the love

AWS has added five new generative AI solutions to its portfolio. For example, Amazon Bedrock is now generally available.

Of the new generative AI solutions AWS wants to make it easier to build applications with AI functionality. Attention is paid to guaranteeing better security and accessibility to AI models with their own data.

The most important introduction is the general availability of Bedrock. This is AWS’s managed generative AI service. This allows customers to discover, train, and fine-tune their own AI models with their own data on AWS infrastructure. The service provides access to well-known AI models such as those of AI21 Labs, Anthropic, Cohere, Stability AI and Amazon’s own models.

Access to Llama 2

In addition, Amazon will also give Bedrock access to Meta’s Llama models and in particular the newer Llama 2 model in the very short term. The service would be the first managed AI service to provide access to Meta’s recent AI model via an API.

Managed service Titan Embeddings

Amazon Titan Embeddings capabilities enable users to quickly build AI-based applications from large data sets. This concerns foundation models with large data sets that convert text into numerical data or embeddings. These can be used to improve contextual searches based on semantics.

This functionality is now available as a managed service.

Improvements to CodeWhisperer

Enhancements to the Amazon CodeWhisperer intelligent coding tool provide new customizable capabilities. The tool now builds code from existing code and ensures that good quality code remains available when developers bring tools together. No sensitive data is leaked and no customer data is stored or logged from the customer’s development environment.

More QuickSight functionality

Furthermore, Amazon QuickSight, AWS’s business intelligence service, is receiving improvements with generative AI, in particular the generative tool QuickSight Q. Users can now more easily ask questions with keywords and also indicate more complex ideas.

You might also like