Accelerate your SAP modernization with Kiro – Benchmarks and Use-Cases

In this blog, we will dive-deeper into each use-case mentioned in our launch blog. We will share some of the results partners and customers have achieved using the Kiro based sample agents during their SAP modernization projects. You can download these sample agents here.

Here is a recap of the primary use cases covered in launch blog.

1. Upgrade ABAP code for SAP S/4HANA compliance
2. Refactor ABAP to clean core
3. Modernize from SAP PI/PO to SAP BTP
4. Modernize from SAP Business Warehouse (BW) to SAP Datasphere in Business Data Cloud (BDC)
5. Generate functional and technical specifications
6. Automate unit testing

This benchmarks throughout this document were completed were conducted in an AWS internal development environment. The Kiro credits used for the benchmarks are well within a Kiro Pro subscription.

1. Upgrade ABAP code for SAP S/4HANA compliance

The first use-case is to convert legacy custom SAP ECC ABAP code to SAP S/4HANA-compliant standard code. For enterprises with thousands (or tens of thousands) of custom ABAP programs, this agent automates the most labor-intensive remediation work. Before converting the code, the agent creates a functional specification, a technical design document, and unit test classes. During the conversion, the agent improves the code to conform to your defined enterprise standards. To achieve this, we provide a fully configurable set of over 80 sample rules and best-practices, covering performance standards, coding standards, security standards, documentation standards, and avoiding code duplication. After the conversion, the agent executes the unit tests to ensure the integrity of the business logic.

Benchmark: In an internal AWS benchmark, we achieved an 87% reduction in effort by converting an entire SAP ECC package of 88 objects (5,854 lines of code) to fully SAP S/4HANA-compliant code in 4.5 hours, with all generated unit tests passing. This benchmark used 377 Kiro credits.

NTT Data Global Solutions (GSL) is using the ABAP agent to help customers accelerate migrations to SAP S/4HANA. GSL’s new migration service, Neo i-KOU, shortens the ABAP modification phase by up to 95%. In their product testing, they converted ABAP code that traditionally required more than 4 man-months to convert with only 0.21 man-months of effort.

2. Refactor ABAP to Clean Core

This use case assesses and remediates custom ABAP code for compliance with the Clean Core Extensibility Model, which grades each customization from A to D based on how cleanly it sits outside the core SAP S/4HANA system. Keeping customizations at the highest levels, A and B, is what lets customers modernize faster and cut costs, because they can rapidly upgrade and adopt new SAP core features without breaking their custom code. AWS helps customers get there by assessing their existing ABAP customizations against these standards and remediating the code that falls short.

Benchmark: In an internal benchmark, our agent identified 116 “Extensibility Level D” ABAP findings across 143 objects (10,200 lines of code). With a runtime of 55 minutes, 100% of the Level D findings were remediated. The agent replaced non-standard code with appropriate SAP-released objects or by extending them adhering to clean core principles. It is important to highlight that the automated clean core remediation is limited only to objects where SAP has released standard objects. Most customers should expect to have a percentage of custom code where released objects are not available. The objects will require manual review and a decision on clean core options (i.e. convert to side-by-side, leave as-is, adopt fit-to-standard). As a best practice, we recommend first using the agent to analyze all custom code and create a report showing current Extensibility Levels (A, B, C, D) per ABAP object and identify which are eligible for automated remediation.

3. SAP PI/PO to SAP BTP Integration Suite Migration

This use case focuses on converting SAP PI/PO interfaces to the SAP Business Technology Platform Integration Suite (BTP IS). SAP PI/PO, SAP’s middleware platform, is approaching end of standard maintenance in December 2027. Customers need to migrate dozens to hundreds of interfaces to BTP IS, and manually converting a single interface can take days to weeks depending on its complexity. To address this challenge, we developed a modernization agent. With a single prompt, the agent studies the existing PI/PO configuration, writes the functional specification and technical design, converts PI/PO artifacts to BTP-ready form, and deploys the result to BTP IS, driving significant time and cost savings. In addition to SAP PI/PO migrations, the agent can also develop net-new BTP IS interfaces directly from a requirements document.

Benchmark: In an internal benchmark, our agent converted 22 SAP PI/PO interfaces to the SAP BTP Integration Suite in approximately two hours. These interfaces contained 25 message mappings and spanned a variety of interface types, including IDOC, SOAP, REST, synchronous and asynchronous. The agent read the source communication channels, agreements, and shared function libraries to determine each target adapter and reproduce the mapping logic. It then created the SAP BTP IS interfaces, deployed them against a live tenant, and tested each mapping with unit-level data against the source rules. The agent successfully handled complex requirements such as routing, fan-out, multicast, value-mapping, IDOC segment maps, and stateful orchestration (ccBPM). For security reasons it does not migrate credentials such as passwords and certificates, requiring a manual step before end-to-end testing. The benchmark used ~220 Kiro credits.

4. SAP BW to Datasphere Modernization Agent

This use case is to accelerate SAP BW modernization and automating the work end-to-end. The agent starts by reading all of your existing SAP BW data pipelines and data models, then writes the functional and technical specifications for each one and converts them into a cloud-native SAP Datasphere equivalent. Customers taking the SAP BW Private Cloud Edition (PCE) path can pair the agent with SAP tools such as the Data Product Generator and Query Template Generator.

This addresses another workload facing the same deadline. SAP BW, an enterprise data warehouse platform used by thousands of companies, is also approaching end of standard maintenance in December 2027. SAP encourages customers to move to SAP Datasphere, and there are two paths to get there. Companies can move directly to SAP Datasphere, or they can move first to an interim platform, SAP BW Private Cloud Edition (PCE), which lets them continue running SAP BW beyond 2027 and adopt Datasphere at their own pace.

The agent for SAP BW is a guided workflow built on Kiro Specs, keeping your architects and developers in control at every step. They discover the data flows, map them to SAP BDC Data Products, build the design documents, and generate and deploy the converted Datasphere objects, applying your organization’s own best practices and data strategy along the way.

Benchmark: In an internal benchmark, this agent converted a legacy SAP BW environment containing Sales and Distribution content to SAP Datasphere in 2.2 hours. The SAP BW environment was based on Layered Scalable Architecture (LSA++) architecture comprised of 32 Data Store Objects, 13 Transformations, 6 Composite Providers, and 1,273 fields. The final SAP Datasphere environment included 32 Local Tables, 13 Data Flows, and 6 Analytic Models. The benchmark used 856 Kiro credits.

AWS partners, DXC and Kyndryl, are using the SAP BW agent to help accelerate their customers’ journeys to SAP BDC. DXC is accelerating digital transformation by combining Kiro and Modernization Agents with ‘DXC Fast BDC’ to automate complex assessments, activate AI-powered business insights, and deliver business value in as little as 10 weeks. At Kyndryl, Poshan Ponnamreddy, SAP US Data Lead, stated, “The combination of AWS Modernization Agents, Kiro, and Kyndryl’s Agentic AI Framework is helping customers reduce SAP modernization timelines. AI-powered conversion capabilities automate code analysis and transformation, reducing manual effort by up to 80% while improving quality. Together, these innovations accelerate migration outcomes and significantly improve time-to-value.”

5. Generate functional and technical specifications

We frequently hear about code written “decades ago” without documentation, and the subject-matter experts have long since left the organization. Using AI to analyze code and generate documentation is a well proven use-case that works across a wide range of SAP platforms, including SAP ABAP, PI/PO, BW, and more. The first step of many SAP projects starts by analyzing and documenting legacy code to understand the as-is functional and technical specifications.

The modernization agent built using Kiro sample code can provide capabilities to help you generate documentation in your desired format and according to your company’s documentation standards. You can also identify unused code, which can avoid unnecessary code modernization effort.

Toyota Chile is migrating from SAP ECC to SAP S/4HANA and lacks documentation and understanding across hundreds of legacy SAP ABAP objects. Rolando Sabatino, Senior Software Engineer, shared his experience stating, “Kiro with AWS’s Modernization Agents for SAP on AWS is allowing us to simplify and reduce the project effort. We used Kiro to identify unused code and obsolete tables which we excluded from the migration scope. We also generated functional documentation for business users and technical specifications for developers. The documentation was generated in hours, avoiding months of manual effort.”

6. Automate Unit Testing

Unit testing is a lesser-known but impactful approach to confirming functionality, and it is what makes AI code conversion highly accurate and reliable. It significantly minimizes AI hallucination during code modernization, so the results hold up. As a best practice, AWS recommends generating unit tests for legacy code before modernizing it, which gives the AI agents a clear success criterion to validate the converted code against.

It is worth expanding on why this matters so much. As AI coding tools such as Kiro and Claude Code have matured to include more agentic capabilities, they autonomously iterate to solve problems in a way that mimics how humans work. When a human writes code and finds a bug, they enter a cycle of analysis, editing, and re-testing that continues until the bug is resolved. Agentic tools run this same loop and use the unit test to confirm when the issues are fixed. Put another way, the unit test tells the agent what success looks like, letting it cycle through analysis, editing, and re-testing until no errors remain. The inclusion of automated unit testing is one of the main reasons our modernization agents hit the high-quality benchmarks mentioned earlier.

Safeguards for Accurate, High-Quality Code

A common and valid industry concern is the accuracy and quality of AI-generated code. Customers frequently ask us, “how can enterprises confidently apply AI to their most critical workloads such as SAP?” Accuracy and quality risks are amplified in ABAP, where code frequently references logic and libraries housed in other ABAP objects. Most AI coding assistants cannot fetch that external, dependent code, which often causes hallucinating. AWS takes accuracy and quality seriously and has incorporated unique safeguards into these sample agents.

The agents built leveraging these sample agents pull in all dependent code and libraries, ensuring Kiro has the full context. This is critical for generating accurate code and maintaining business logic during conversions. Second, customizable steering files let you define your own coding, performance, security, and documentation standards. The sample agents ship with sample steering files containing over 80 rules to get you started. Third, our unique approach to unit testing creates and tests code before conversion, then re-tests the exact same business logic after, significantly reducing improving quality and saving time.

Auditability is another important safeguard. Modernization agents built using these Kiro based sample agents can save conversation history and log decisions, aiding debugging and fine-tuning. Notably, two customers reviewed their AI automation approaches with their Internal Controls and audit teams. Contrary to what they expected, the auditors preferred the transparency of decision logging, since it exceeds the documentation most employees maintain by hand.

While these agents automate many parts of the software development lifecycle, it is important that customers still perform thorough, end-to-end testing and business validation before moving it to production.

Conclusion

SAP modernization has remained one of the most complex and resource intensive undertakings in the IT enterprise. By sharing the sample agents, we aim to provide our customers and partners a more accelerated and affordable path to modernization. You can download these sample agents here and find more information here.