Clean S/4 intelligence

The era of monolithic custom code is coming to an irrevocable end, and for existing SAP customers who are still deeply rooted in the world of ERP/ECC 6.0, this is an existential turning point. While the marketing drums from Walldorf are tirelessly preaching SAP Clean Core as the new gospel, many companies are stunned by millions of lines of historically grown Abap code slumbering like a technological fossil in the digital underground of their systems.

This tension between the need for modernization and the burden of the past is precisely where the start-up Nova Intelligence from San Francisco (USA) comes in with an agentic AI platform that promises to solve the dilemma of technical debt not through laborious manual work, but through artificial intelligence.

Agentic AI platform

„Nova Intelligence systematically reduces SAP customer code to the minimum that is actually required for a company's differentiating business processes,“ explains Nova co-founder Professor Alexander Zeier. „Our Agentic AI platform analyzes the complete customer-specific custom code of an SAP system, reconstructs the functional logic and generates documentation at specification level. Each custom code module is broken down into its business functions and systematically evaluated against the S/4 Hana standard - with clear recommendations and comprehensible reasons as to whether a function can be replaced by the standard or requires a customer-specific enhancement.“

At the heart of this approach is a fundamental departure from traditional migration paths. Until now, existing SAP customers often only had the choice between plague and cholera: the brownfield approach, where legacy systems are simply taken along and complexity is cemented, or the greenfield approach, which is clean but destroys the process knowledge built up over decades. Nova Intelligence positions itself somewhere in between with an intelligent clean-field concept.

The AI system does not act as a mere translation tool for code, but as an archaeological architect. The platform's AI agents delve deep into the old SAP Z namespace, not only analyzing the syntax, but also reconstructing the business logic (business intent) behind the code and documenting it at a level that resembles a professional specification. This is the decisive lever for S/4 conversion: instead of blindly migrating code, Nova enables a well-founded fit-to-standard analysis. The AI checks whether the former special lock can now be mapped by a modern S/4 standard process and thus eliminates unnecessary ballast before it can clog up the new system.

Nova reduces SAP customer code to the minimum that is always required for the differentiating business processes.

Prof. Dr. Alexander Zeier,
Co-Founder & Chief Scientist,
Nova Intelligence

„Customer code whose functionality is already included in the S/4 Hana standard is consistently identified and replaced by the standard - this immediately reduces the custom code inventory,“ Professor Zeier knows from many proof-of-concepts. „Only the extensions that are really needed are completely redeveloped on the basis of these specifications in a clean-core-compliant manner. At the same time, the same Nova platform also supports the implementation of new technical requirements - not just the modernization of existing code.“

The result: SAP teams typically work up to twice as fast, the risk of SAP modernization is significantly reduced and the total cost of ownership is sustainably lowered. „In combination with DMI JiVS, which structurally decouples data and makes it manageable, a lean, upgrade-stable and sovereign S/4 Hana system is created,“ says Alexander Zeier, praising the successful collaboration between Nova Intelligence and the Swiss SAP partner Data Migration International, see also E3 cover story March 2026.

Clean core and data management

The strategic relationship between Clean Core, Agentic AI and clean data management by DMI JiVS can be described as a symbiosis of compulsion and liberation for the existing SAP customer. Critically analyzed, this relationship is not optional, but existential: Clean Core is the technical life insurance for the use of AI, while AI agents are paradoxically the only economically viable tool for achieving this clean state at all. Conclusion for the existing SAP customer: Clean core by Nova Intelligence and clean data management by DMI JiVS are not an IT end in themselves for nicer architecture, but the hard business requirement to remain capable of acting in the age of AI. Without a clean core, AI agents remain an incalculable risk. It is strategically wise to use AI agents now to establish the clean core before attempting to scale AI agents for operational business processes.

The AI build component of the Nova Intelligence platform is used for the remaining, competition-differentiating enhancements. This demonstrates the technical sophistication in dealing with the Business Technology Platform (SAP BTP). Nova Intelligence automatically generates new, clean-core-compliant code that strictly separates the stable SAP core from the flexible extension layer. The system is a perfect master of the modern SAP architecture: it can generate on-stack extensions using Abap Cloud (embedded steampunk) as well as implement decoupled side-by-side applications on the BTP.

Migration and data hygiene

The S/4 migration cockpit is massively relieved. Nova may not be an ETL tool that shovels data, but it cleans up the blueprint of the target system so fundamentally that the migration cockpit only has to migrate to clean, standardized structures instead of a mess of old Z-tables. It is the strategic vanguard that cleans up the battlefield before the actual data migration begins. The „data shoveling“ can be taken over in a transformation scenario by JiVS from the Swiss SAP partner DMI.

Founder and Group CIO Thomas Failer from DMI puts it in a nutshell: „If the foundation is fragile, cracks will appear in the walls after a short time.“ When agents access inconsistent data in a modified system, incorrect decisions are automated and accelerated. The clean core enforces the data hygiene that is essential for reliable AI results (grounding). AI becomes the architect of the clean core (nova effect). Many existing customers are faced with millions of lines of legacy code that would be prohibitively expensive to clean up manually. AI agents become the exit strategy from this technical debt.

The historical arc that Professor Dr. Alexander Zeier spans is interesting here. As co-inventor of the SAP Hana database alongside Professor Hasso Plattner, he knows the architectural DNA of the SAP ERP system better than almost anyone else. His role at Nova Intelligence is not that of an aloof figurehead, but that of a Chief Scientist who takes responsibility for the consequences of his own invention. Zeier recognized that the introduction of Hana and S/4 brought enormous performance advantages, but left existing SAP customers alone with the adaptation of their old Abap code.

Nova Intelligence is therefore an answer to the complexity problem triggered by the in-memory revolution.

Complexity problem with Hana

The relationship with SAP is symbiotic, but independent: While SAP provides the standard, Nova provides the key to being able to use this standard at all without giving up its own identity. IT platforms such as Nova Intelligence use Agentic AI to not simply copy old SAP custom code (Z-code), but to extract its technical business logic and document it at specification level.

AI agents compare the extracted logic with the current S/4 standard. They identify where the standard can replace the old code and thus reduce the custom code to the business-critical minimum. For the remaining necessary extensions, AI agents (SAP AI Build) automatically generate new, clean-core-compliant code - whether for the Abap Cloud Environment or as a side-by-side extension on SAP BTP. This strategy brings together Nova and SAP elements on the SAP Business Technology Platform. This is also where the AI services (SAP AI Core, GenAI Hub) live, on the BTP. The BTP acts as a link: it keeps the ERP core clean by hosting the AI agents and extensions and connecting them to the ERP data via clean APIs.

Critically, Nova Intelligence breaks with the old computer science paradigm of „algorithms and data structures“, as taught by Professor Niklaus Wirth at ETH Zurich. In the old world, humans had to laboriously define the algorithms in order to manipulate the data structures. Nova turns this around: AI learns the implicit algorithms (the business logic) from the existing data and structures and generates new, more efficient structures from them. It is the transition from deterministic programming to probabilistic programming.

Where monolithic Z programs used to interrupt the process flow, decoupled, API-based services now ensure a seamless process that fits in with the vision of the Intelligent Enterprise.

Courage for transparency

The advantages for existing customers are obvious, but they require the courage to be transparent. Nova mercilessly reveals where spaghetti code has been produced in the past. But it is precisely this honesty that is the only way to clean core.

The system reduces technical debt not cosmetically but structurally, lowers the total cost of ownership (TCO) and makes the company capable of innovation again.

At a time when SAP is increasingly making many innovations available exclusively in the cloud and on the BTP, Nova Intelli-gence is the vehicle that enables customers to jump on this train instead of being left behind on the platform of the outdated ECC world.