Amazon Web Services (AWS) has released a collection of AI agents built on Kiro to speed up enterprise SAP modernization. The tools are designed to analyze and convert ABAP code, update middleware integrations, ease SAP Business Warehouse migrations, and generate automated tests. According to an internal AWS benchmark, the agents can cut the effort needed to modernize existing ABAP code by up to 87%, while some of the company’s partners say they have achieved reductions of up to 95% on specific conversion tasks.
Kiro and SAP modernization in 30 seconds
- AWS has released sample agents built on Kiro to modernize custom development and legacy SAP systems.
- The agents can analyze, convert and test ABAP code, as well as generate technical and functional documentation.
- AWS reports an effort reduction of up to 87% in an internal ABAP modernization benchmark.
- NTT Data Global Solutions says it has cut conversion effort by up to 95% with its Neo i-KOU service.
- The initiative also covers migrations from SAP PI/PO and SAP Business Warehouse, whose standard support end dates are approaching.
The proposal aims to reduce the manual work required by migrations of enterprise systems with decades of customizations and integrations. Many organizations run financial, logistics and analytics processes on custom-built code that can’t be replaced without first checking that it preserves their business rules. AWS wants to use agents capable of working across large codebases to speed up that transition without giving up testing and technical controls.
Kiro automates ABAP conversion and SAP documentation
Kiro is AWS’s AI software engineering agent, following the frontier agents the company unveiled to transform how software gets built. AWS is positioning Kiro for use in complex modernization projects, where teams need to analyze numerous components, identify dependencies, and coordinate changes across different parts of an application.
The released agents target several tasks. One is adapting ABAP code from SAP ERP Central Component (SAP ECC) systems to meet SAP S/4HANA requirements. They can also refactor custom development to move it closer to SAP’s Clean Core model, which aims to reduce modifications to the standard core and favor the use of compatible interfaces and APIs.
The automation isn’t limited to transforming code. The agents can analyze existing programs and configurations to generate technical and functional specifications, document business requirements, and create unit tests. The latter run before and after the conversion to check that business logic keeps behaving as expected.
AWS says its agents can process hundreds of ABAP objects from a single instruction. The company also states that, in an internal benchmark, the process of analyzing, converting and testing legacy ABAP code cut the required effort by up to 87%. The figure comes from an internal evaluation and should not be read as a guaranteed improvement for any given SAP project.
Complexity depends on factors such as the volume of custom code, dependencies, documentation quality, and available tests. An application with numerous specific modifications may need more human intervention than a relatively simple development.
SAP PI/PO and Business Warehouse migrations ahead of December 2027
Another part of the initiative focuses on middleware and data platforms. AWS has designed agents to help migrate SAP Process Integration and SAP Process Orchestration (SAP PI/PO) interfaces to SAP Integration Suite, within SAP Business Technology Platform (SAP BTP).
These interfaces connect SAP systems with external applications, business partners and other enterprise services. Migrating them can require reviewing message transformations, configurations, connections and dependencies. AWS aims to automate part of that analysis and conversion to cut the time needed to rebuild integrations on the target platform.
The agents are also aimed at modernizing SAP Business Warehouse (SAP BW) toward SAP Datasphere, within SAP Business Data Cloud. Here, the goal is to move analytics and data management processes to a newer architecture while preserving the organization’s information needs.
AWS frames both workstreams around a key date: December 2027, when standard support for SAP PI/PO and SAP BW is set to end, according to published information. The company also cites the end of standard support for on-premises SAP ECC as another reason to plan migrations, though this announcement doesn’t specify a single date applicable to every customer and scenario.
The deadline could prove especially demanding for organizations that depend on hundreds of interfaces or maintain systems with extensive customizations. Automating part of the inventory, documentation and conversion work can help prioritize tasks, but it doesn’t remove the need to validate results in each environment.
AWS highlights results from Toyota Chile, Adobe and its partners
The company cites tests run by customers and consulting firms. Rolando Sabatino, a senior software engineer at Toyota Chile, used the agents to analyze and document hundreds of custom ABAP objects. According to the testimonial published by AWS, the analysis made it possible to identify obsolete code and narrow the scope of the migration.
Alex Kakhanovich, an SAP enterprise architect at Adobe, evaluated Kiro ahead of a large-scale migration to RISE with SAP and highlighted its potential to speed up correction and adaptation tasks. The announcement does not provide a specific time-savings figure for Adobe.
Partners incorporating the agents into their migration tools include Accenture, Capgemini, Deloitte, DXC, Kyndryl and NTT Data Global Solutions.
NTT Data Global Solutions reports an effort reduction of up to 95% in ABAP conversion within its new Neo i-KOU migration service. Capgemini, for its part, says its new Smart Integration AI Agent, built using Kiro’s sample agents, can speed up the cycle from analysis to integration building by 40% to 60%.
Deloitte is also working with AWS on Kiro-assisted migration approaches to move SAP PI/PO environments to SAP BTP Integration Suite. In this case, the announcement highlights reduced effort and delivery times but doesn’t include a specific percentage.
These figures come from different experiences and services, so they aren’t directly comparable. The 87% reported by AWS comes from an internal benchmark, while the 95% figure and the 40%-to-60% range are attributed to specific partner services. Results will depend on the task, the starting environment, and the methodology used to measure effort.
Tests, coding rules and traceability to keep changes under control
One of the risks of using AI agents on enterprise applications is that a seemingly correct conversion could alter a poorly documented business rule. In SAP, a change can affect billing, inventory, procurement or financial processes, so generating valid code isn’t enough to prove the application works the way it did before.
AWS says the agents include customizable configuration files with more than 80 predefined rules covering coding, security and documentation standards. They also aim to retrieve ABAP code and dependent libraries so the model has more context during conversion.
The system runs unit tests before and after the changes, along with a conversation history and a decision log to support auditing. AWS says some customer audit teams consider these logs more thorough than the documentation they typically maintain manually.
These measures aim to reduce errors and hallucinations, but they don’t remove the need for technical review. Unit tests only verify the scenarios they cover, so complex migrations may require integration testing, functional validation, and checks with business users.
The agents are available through a registration process that grants access to a GitHub repository with the sample code and documentation. Companies can request guidance from their AWS account team or work with an AWS SAP Competency partner to plan their project.
The initiative turns agentic AI into a tool for tackling specific enterprise modernization tasks. Its value will depend on how much code and integration it can reliably transform, the quality of the testing, and how much human effort remains necessary before rolling out the changes in production.
Frequently Asked Questions
What is AWS’s Kiro for SAP?
Kiro is an AI software engineering agent. AWS has released sample agents built on Kiro to analyze, convert, document and test custom SAP code and integrations.
How much can it reduce SAP modernization work?
AWS reports an effort reduction of up to 87% in an internal ABAP modernization benchmark. NTT Data Global Solutions says it has reached up to 95% in conversion with its Neo i-KOU service, though these results come from different scopes.
Which SAP systems can these agents modernize?
Use cases include ABAP code from SAP ECC to SAP S/4HANA, refactoring for Clean Core, migrating SAP PI/PO to SAP Integration Suite, and modernizing SAP BW toward SAP Datasphere.
How can Kiro’s agents be tested?
AWS offers access through a registration form. After completing the process, participants can access the GitHub repository with the sample agents and their documentation.
