ELAU MC-4/11/10/400
Specifically, Amazon Cloud Technology is launching a new generative AI infrastructure consisting of three layers of technology stack: infrastructure layer, basic model service layer, and AI application layer. The first layer of the stack is an innovation in storage and computing, belonging to the underlying infrastructure, including the cloud AI chip Amazon Traineum2 designed for machine learning training, and the fourth generation self-developed server CPU chip Amazon Graviton4. The second layer of the stack, Amazon Bedlock platform, and the third layer of Amazon Q reduce the cost of developing and using generative AI for enterprise users from the basic model service layer and AI application layer, respectively.
Amazon Bedlock services provide a “fast track” for generative AI development for enterprise users. In the process of developing a large language model, the model itself is not good at executing specific tasks across company systems and data sources. Enterprise developers must write code to coordinate the interaction between the model, system, and users, so that the application can execute a series of API calls in logical order. At the same time, data security and privacy policies need to be formulated, which are time-consuming and require professional knowledge, It will increase the time cost of developing generative AI applications. Amazon Bedlock standardizes these complex processes, allowing enterprise users to deploy without even writing a single line of code. At the same time, enterprise users can directly utilize Amazon Bedlock to access the latest versions of almost all industry-leading models such as Amazon Titan Family, Anthropic Claude 2.1, Meta Llama 2 70B, and Stability AI Stable Diffusion XL 1.0, as well as the wide range of features required to build generative AI.
At the application layer, Amazon Cloud Technology has launched the enterprise level generative AI assistant Amazon Q. With the help of Amazon Q, enterprise users do not need to search for various business-related data from the vast sea of work documents, but can directly achieve it through inquiries. For example, developers can maintain their applications by querying Amazon Q for code from other engineers.
Meanwhile, Amazon Q has great flexibility. Currently, developers often need to invest a lot of effort and work to keep up with the speed of generative AI technology iteration, quickly design and deliver new features, manage the end-to-end lifecycle of applications and workloads, and balance priorities between maintaining existing products and building new features. Amazon Q fully supports customization based on customer business to help enterprise developers focus on development itself.
In short, Amazon Cloud Technology provides out of the box tools and services for enterprises who want to develop generative AI applications but face various challenges through multiple products and services, including machine learning, big data analysis, and application development environments, standardizing and modularizing the originally complex development of generative AI applications. As long as enterprises have the willingness to develop generative AI applications, various complex technical problems are no longer obstacles to their actions.