Amazon releases Bedrock, launching multiple new features to help enterprises leverage generative AI technology

Amazon Web Services recently launched five new generative artificial intelligence products in the ecosystem. These products will help enterprise customers use their own data to build artificial intelligence applications and Provides better security and model accessibility.

These new services include the general launch of Amazon Bedrock, Meta Platforms Inc.’s Llama 2, Amazon Titan embedded AI, Amazon CodeWhisperer’s new coding capabilities, and QuickSight’s generative AI enhancements for business intelligence

Swami Sivasubramanian, vice president of data and artificial intelligence at AWS, said: "The explosion of data, the use of scalable computing, and advances in machine learning have led to a surge in interest in generative artificial intelligence over the past year, inspiring the potential for change. entire industries and reimagine new ideas for how work gets done.” “Today’s announcement is a major milestone in making generative AI within reach, from startups to large enterprises, from developers to data analysts, in any industry. Any employee can use generative AI.”

Amazon Bedrock is the company’s fully managed service for generative AI, providing access to underlying models. Customers can use the service to discover, train and tune their own models using their own proprietary data in the secure environment of Amazon's high-performance infrastructure without having to worry about managing the rest.

Customers now have immediate access to high-performance models from leading AI companies such as AI21 Labs, Anthropic, Cohere Inc., Meta Platforms Inc., Stability AI Ltd., and custom models from Amazon. The service also offers a host of features that customers can use to build their own AI applications that can talk to customers, aggregate documents, generate images, and provide AI-driven searches.

Amazon claims that Bedrock is the first fully managed service to provide Meta’s Llama 2 large-scale language model, implemented through an application programming interface. In the coming weeks, developers will be able to build artificial intelligence applications using Llama 2, which is optimized for AWS infrastructure by Bedrock (based on 13 billion and 70 billion parameter models)

Titan Embeddings Now generally available, it allows customers to quickly create artificial intelligence applications based on large data sets. The Amazon Titan base model is a family of models based on large datasets that convert text into numerical representations called embeddings, which can then be used for semantically-based context-enhanced search. This can be used to enhance AI-driven searches, provide better personalization, and other use cases.

Because building an embedding model requires a large amount of data and rich machine learning expertise, many companies have difficulty implementing this function. Now, with Titan Embeddings, enterprise customers can easily achieve this functionality through managed services. This feature on AWS supports 25+ languages and has a context length of 8,192 tokens, which means it can handle anything from single words to extremely long documents

Amazon CodeWhisperer is an AI coding companion for developers , working with developers by suggesting code snippets, rewriting code, and explaining code. Its models are trained on billions of lines of open source code, which enables it to provide these capabilities to developers. Amazon has upgraded it so that enterprise customers can customize it using private code from an internal code base so that it can provide recommendations based on a business's unique needs.

Prior to the update, developers might use the tool to help them write generic code, but CodeWhisperer had no knowledge of a company's specific internal needs or coding requirements. Developers may still spend hours reviewing previously written code to ensure it meets the company's requirements.

With new customization capabilities, CodeWhisperer unifies developer tools and works with developers to maintain already high-quality code based on existing code. Amazon says it will do this in a way that does not reveal confidential information and that no customer information will be stored or logged from the customer's development environment.

"We are using Amazon CodeWhisperer to equip our more than 16,000 engineering staff to build and deliver industry applications faster and more securely," said Pandurang Kamat, chief technology officer of Persistent Systems Ltd. "Some teams Already starting to leverage CodeWhisperer’s new customization capabilities to maximize the value of generative AI-driven code recommendations, we’re already seeing good results."

Latest research finds that Persistent is working with AWS , developers complete coding tasks an average of 28% faster

Enterprise users gain access to new generative AI capabilities through Amazon QuickSight, a unified business intelligence service built on the cloud that can Provides interactive dashboards, reports, and embedded analytics. QuickSight has the ability to perform natural language queries using QuickSight Q, allowing any user to simply type a structured question and get results.

Amazon is introducing new capabilities to QuickSight Q, expanding its natural language capabilities so analysts can more loosely use language to obtain information and insights from the analytics engine. Before the update, it asked analysts to have clearly structured questions like "What are the top ten products in Arizona?" But with the power of generative AI, it can now handle far more complex ideas.

Now analysts can get the visualizations they want simply by describing the results they want, and ask follow-up questions to refine complex calculations based on the generated reports. Once you're done, answers and charts can be quickly added to a panel or report with just one click.

For example, analysts can now ask AI to create a visual chart for "Monthly trends in citrus sales in 2022 and 2023" and AI will choose the chart format that best meets the request, such as a line chart or a bar chart . If the analyst prefers a different chat format, they can later ask to change it to a spreadsheet.

If the question is unclear, queries can match multiple data fields, and new generative AI capabilities also allow QuickSight Q to provide analysts with relevant questions

The above is the detailed content of Amazon releases Bedrock, launching multiple new features to help enterprises leverage generative AI technology. For more information, please follow other related articles on the PHP Chinese website!