AWS App Studio: Using GenAI for enterprise applications

Contents

Amazon calls its new service App Studio, which allows users to create business applications using generative AI. AWS is explicitly gearing the service towards regular users and not developers – whereby users should certainly have a technical understanding, but do not need to have any programming experience. The finished web application should be ready for use in just a few minutes.

First GenAI, then drag-and-drop and low code

To get started, a data source is selected: This can be an AWS service; connectors are already available for S3, Aurora, DynamoDB & Co. The same applies to Salesforce; according to the announcement, App Studio also integrates hundreds of third-party services via API. Managers can set up connectors for their employees in advance, which are then available to the entire group. Users should not have to worry about how these connections work – the AI should take care of the integration.

As with other GenAI services, users then enter a prompt in App Studio that describes the finished application. The system then translates this input into a detailed description that can be edited further. App Studio then generates an initial demo. This takes the user to a GUI in which they can adjust the structure of the program using drag-and-drop. Additional input and output fields can also be added here.

Users can also further customize the data sources and the storage location for the application data. Finally, App Studio also provides an overview of the application's logic, which can also be edited further. No programming knowledge is required for this either: AWS instead offers a low-code environment where users can drag and drop individual elements into the flow of the program and adapt them to their own needs using simple commands.

App Studio versus Excel spreadsheets

Once you have completed your application, there are three further steps: In a development environment, it is in a sandbox; it is not yet connected to the actual data sources at this point. Instead, App Studio generates suitable data so that users can test their application at this stage. The subsequent testing phase accesses the set data sources; employees can try out the application at this point and provide feedback. Once this is complete, the application is published in the production environment.

In its examples, however, AWS also shows that it is not enough to simply describe an arbitrary application to the AI via a prompt – knowledge of the company's own technical infrastructure is required. Accordingly, Amazon is not gearing its new App Studio towards regular company users; rather, project managers are to replace processes that previously used Excel spreadsheets, for example, with such applications.

Obviously, this goal is not new: low-code tools make the same promise. However, AWS sees them as too restrictive. The barriers to entry are too high, users have to acquire platform-specific knowledge, and IT managers usually prevent such applications for security reasons. It is not clear from the announcement why this should be different with App Studio.

Frequently used applications will be more expensive

As with any cloud service, the question of fees remains: Here, Amazon is relying on a model typical of the cloud - customers pay for what they actually use. Accordingly, the creation of business applications via App Studio is free of charge; instead, AWS charges for the time users spend in the application. This should save customers up to 80 percent in costs compared to low-code tools. Amazon does not provide any information on how this was calculated.

App Studio is currently in a preview phase in the US West region (Oregon). AWS has already made the GenAI service available to larger companies and institutions in advance. More details can be found in the announcement on Amazon.

(fo)