Amazon Web Services Inc. today debuted a set of artificial intelligence agents that will help streamline the work of its professional services group.

The group is known as AWS Professional Services, or ProServe for short. It helps customers move legacy workloads to the cloud giant’s platform. ProServe also builds greenfield applications using AWS services.

According to the Amazon.com Inc. unit, the new agents will help compress tasks that usually take several months into days. They are also expected to reduce project costs in the process. AWS has already used the agents in multiple client projects.

The agent collection is headlined by the AWS Professional Services Delivery Agent. It enables customers to launch a project by uploading diagrams that describe the architecture of the application they wish to build. The AI can also extract project requirements from meeting notes, documentation and other files.

Before work begins on a project, the service creates a proposal for the client along with a statement of the work. The latter document contains details such as the amount of time the project is expected to take. In some software modernization initiatives, manually drafting those documents can take weeks. AWS says the Professional Services Delivery Agent performs the task in a few hours.

After a project kicks off, the tool uses a collection of specialized agents to carry out technical work. Those agents generate code, test that the code works as expected and deploy it. The process is supervised by a ProServe consultant who ensures that the outputted code is aligned with the customer’s requirements.

If the Professional Services Delivery Agent receives a request to move a legacy workload to the cloud, it assigns the task to an agent based on AWS Transform. The latter service, which launched in May, uses AI to automate software migrations. It can move COBOL-based mainframe programs, VMware workloads and .NET applications to AWS.

One of the biggest challenges involved in moving software to a new environment is adapting its dependencies. A dependency is a software module on which a program relies to work. If a dependency can’t be migrated together with the application it powers, it has to be replaced with a different software module. That usually involves a significant amount of manual coding.

According to AWS, the AI agent it built on Transform automates much of the workflow. It maps out the dependencies that an application uses and creates a plan for moving the software to the Amazon unit’s cloud. It then generates the necessary code. 

“This agent incorporates a knowledge base of learnings from thousands of migrations AWS ProServe has completed and automation capabilities from our professional services Cloud Migration Factory to accelerate project delivery,” Francessca Vasquez, vice president of AWS professional services and agentic AI, wrote in a blog post.

The Amazon unit’s top rivals in the cloud are also using AI agents to speed up software migration projects.

Google LLC is currently in the process of moving its workloads to its custom Axion series of central processing units. Last month, the search giant disclosed that it has built an AI agent called CogniPort to automate the task. The tool not only rewrites code to run on Axion chips but also generate automated tests that check the code for bugs.

Microsoft Corp., meanwhile, provides application modernization features as part of its GitHub Copilot programming assistant. Those features focus on helping companies migrate .NET applications to Azure. GitHub Copilot can also move programs that use legacy .NET versions to the latest release of the Microsoft-developed application framework. 

Photo: AWS