SAP BTP · Clean Core
Clean Core BTP Extension Strategy
Every classic Z-program that stays in the core — built on unreleased objects or modifying SAP standard — is technical debt that blocks AI. The extension strategy determines where each capability lives — and how it connects back without breaking your upgrade path.
Clean core is DEBCOR's practice area — not a methodology checkbox. We classify your custom code footprint, architect the extension model, and build the BTP extensions and RAP objects that move each Z-program off the core. We also deploy the AI agents that clean core makes possible.
Not sure where your clean core posture stands?
DEBCOR's free AI Readiness Assessment evaluates your landscape immediately — custom code footprint, BTP foundation readiness, data quality, and AI readiness score. 15 minutes. No sales pitch. The clean core analysis is built in.
The extension model
Three Tiers. Every Extension Has a Home.
The extension strategy classifies every custom requirement into one of three tiers. The classification drives the build approach, the governance model, and the timeline.
Side-by-Side Extensions on BTP
Custom applications, AI agents, and integrations built on SAP BTP — outside the S/4HANA core entirely. Technology choice is open: Java, Node.js, Python, LangGraph, n8n. Connects to S/4HANA via APIs and events. Zero impact on the core upgrade path.
- Custom Joule agents and AI automation
- BTP-native applications and UIs
- Integration Suite iFlows and event-driven workflows
- Payment and commerce extensions
DEBCOR's primary BTP engineering practice — where most new capability lives.
RAP/ABAP Extensions — On-Stack & Side-by-Side
RAP (RESTful Application Programming Model) is SAP's programming model for building ABAP services, and where you run it is the architectural decision. On-stack — ABAP Cloud developer extensibility inside S/4HANA: direct access to business objects, released APIs only, upgrade-stable. This is where BAdI implementations and extensions that need to sit close to the data belong. Side-by-side — the ABAP Environment on SAP BTP: a separate system that reaches S/4HANA through released APIs and events, with independent scaling, its own lifecycle, and its own release cadence. Same programming model. Different deployment, different constraints.
- On-stack: BAdI implementations and SAP-standard extension points
- On-stack: ABAP Cloud developer extensibility inside S/4HANA
- Side-by-side: ABAP services on the BTP ABAP Environment
- Z-program replacements in clean-core ABAP
DEBCOR's founder wrote the first ABAP textbook. 30 years of ABAP depth, now applied to RAP.
Z-Program Remediation
Existing Z-programs classified against three outcomes: retire (standard SAP now covers it), remediate (rewrite to upgrade-safe ABAP without BTP migration), or migrate (rebuild as Tier 1 or Tier 2 extension). This classification is one of the most time-sensitive pre-migration activities.
- Custom code impact analysis
- ABAP cleanup and modernisation
- Z-program retirement where SAP standard applies
- Pre-migration custom code strategy
DEBCOR's Custom Code Remediation practice is a dedicated service — not bundled into migration.
Clean core and AI readiness
Clean Core Is the Prerequisite for SAP's AI Roadmap
Every capability SAP announced at Sapphire 2026 — Joule agents, the Autonomous Enterprise, the Knowledge Graph, AI Agent Hub — runs on BTP and requires a governed, well-architected BTP foundation to deploy against. Organizations with Z-programs still modifying the core cannot adopt SAP's AI roadmap on SAP's own timeline. The extension strategy is not a future-state aspiration — it's the current blocker.
Clean core → AI agents deploy faster
When extensions live on BTP with clean API and event interfaces, deploying a new Joule agent or AI automation is a BTP build — not an S/4HANA modification. The extension framework, data access patterns, and governance architecture are already in place. New AI capabilities slot in without a transport or core change.
AI agents accelerate clean core migration
DEBCOR uses AI to accelerate the clean core migration itself — AI-assisted custom code analysis classifies Z-programs faster, AI data cleansing prepares master data for the extension model, and AI-powered migration cockpit tooling compresses the timeline from classification to go-live. The two programs reinforce each other.
Where clean core programs stall
The Z-Program Problem Has to Be Solved Before Cutover
Custom code is the most common reason S/4HANA migrations stall — not because the migration is complex, but because the custom code scope was never properly classified. DEBCOR's clean core approach starts the Z-program analysis before the migration plan is finalized.
~30%
Retire
SAP S/4HANA standard now covers the functionality. The Z-program is retired — no rebuild required. Often the largest category, and the fastest win.
~40%
Remediate
Rewrite to upgrade-safe ABAP without moving to BTP, using clean-core ABAP patterns and extension point adoption. Existing modifications are remediated separately (SPAU/SPDD adjustment) during an upgrade or conversion.
~30%
Migrate to BTP
Rebuild as a side-by-side BTP extension or RAP object. These are the Z-programs that represent genuine custom capability worth preserving as a first-class extension.
Ranges drawn from 10 engagements, 2016–2026, across landscapes of 1,000 to 5,000 custom objects. Actual distribution varies by landscape age and customization depth.
Frequently Asked Questions
What is clean core BTP extension strategy?
Clean core BTP extension strategy is the architectural discipline of extending SAP S/4HANA without modifying the digital core. Every customization, integration, and custom application lives outside the core — on SAP BTP as a side-by-side extension, as a clean-core RAP/ABAP object, or as a BTP-native AI agent — connected via APIs and events. This preserves the S/4HANA upgrade path and is the prerequisite for SAP's AI roadmap.
How do I know if my SAP landscape is clean-core ready?
Use DEBCOR's Clean Core Readiness Assessment at debcor.com/assess/clean-core — a structured self-assessment covering custom code volume, integration complexity, BTP foundation readiness, and migration timeline. Results in 5 minutes. It identifies where your landscape sits across the three extension tiers and what needs to move before S/4HANA cutover or AI agent deployment.
What is the difference between on-stack and side-by-side extensions — and where does RAP fit?
The real architectural choice is on-stack versus side-by-side, not side-by-side versus RAP. RAP (RESTful Application Programming Model) is SAP's programming model for building ABAP services, and it runs in both places. On-stack means ABAP Cloud developer extensibility inside S/4HANA — direct access to business objects, released APIs only, upgrade-stable, and where BAdI implementations and extensions that must sit close to the data belong. Side-by-side means a separate system on SAP BTP that reaches S/4HANA through released APIs and events — with independent scaling, its own lifecycle, and its own release cadence. Same programming model; different deployment, different constraints.
How do AI agents fit into a clean core extension strategy?
AI agents are the most important clean-core extension category in SAP's current roadmap. Joule agents, custom AI agents, and automation workflows all run on BTP — outside the S/4HANA core, connected via APIs and events. A clean core extension strategy that accounts for AI from the start deploys agents significantly faster: the extension framework, data access patterns, and governance architecture are already in place.
What happens to existing Z-programs in a clean core migration?
Z-programs fall into three categories: retire (functionality now standard in S/4HANA), remediate (rewrite to be upgrade-safe without BTP migration), or migrate to BTP (rebuild as side-by-side extension or RAP object). DEBCOR classifies every Z-program before migration and builds the extension strategy for those that must move. This classification should start before the migration plan is finalized.
What is the difference between retiring, remediating, and migrating a Z-program?
When classifying Z-programs for a clean core migration, every object falls into one of three categories. Retire: the functionality now exists in standard S/4HANA — the Z-program is simply decommissioned, which reduces maintenance burden and upgrade risk. Remediate: the Z-program contains genuine custom logic worth keeping, but it needs to be rewritten using upgrade-safe ABAP patterns (extension points, BADIs, enhancement spots) rather than core modifications. Migrate to BTP: the Z-program represents a capability that belongs as a side-by-side BTP extension or RAP object — rebuilt outside the core and connected via APIs. Getting this classification right before migration is what separates a clean cutover from one that carries technical debt forward.
What is SAP's ABAP Test Cockpit and when should we run it?
SAP's ABAP Test Cockpit (ATC) scans your ABAP custom code against the S/4HANA compatibility rules — flagging every object that needs to change before migration can proceed safely. SAP recommends ATC. We treat it as mandatory — we won't scope a remediation without it, because without it the scope is undefined and the timeline is a guess. The analysis should run before the migration plan is finalized, not after. DEBCOR can run and interpret the analysis as a standalone engagement, and can use AI-assisted classification to compress the review cycle significantly.
What is the difference between a user exit, an enhancement, and a Z-program?
Z-programs are custom ABAP programs built entirely outside SAP standard — reports, interfaces, conversion programs. User exits are SAP-provided hooks where custom code is inserted without modifying the standard program itself — when implemented correctly, they are upgrade-safe. Enhancements are broader customizations including BADIs, enhancement spots, and implicit enhancements. Together, all three make up your custom code footprint — and all three need to be inventoried and classified before a clean core migration. Not knowing which user exits and enhancements exist in a landscape is one of the most common sources of migration scope surprises.
How does PI/PO affect AI agent deployment?
AI agents run on SAP BTP — which includes SAP Integration Suite. PI/PO is a standalone product and was never a prerequisite for deploying agents. An agent on BTP reaches S/4HANA through released OData, REST and SOAP APIs via the Destination service and Cloud Connector. If you still run PI/PO for existing interfaces, that only means PI/PO itself cannot serve as the agent connectivity layer — it does not block agents from connecting to SAP processes. Modern agentic connectivity is built on Integration Suite and BTP's released API surface, not on PI/PO.
Start with the free clean core assessment.
DEBCOR's AI Readiness Assessment scores your clean core posture and AI readiness in 15 minutes — no setup, no sales pitch, immediate results. If the score reveals blockers, we can discuss what the extension strategy looks like for your specific landscape.