Custom code isn't the problem — not understanding it is
SAP custom code is not a problem. Not in principle, at least.
Many custom developments have delivered real value to companies for decades. They automate manual work, fill in missing functions and enable processes that the SAP standard alone cannot sensibly cover. Companies, customers, suppliers, users, consultants and developers still benefit from them today.
So why the bad reputation?
There are many reasons. Custom developments cause maintenance effort — even when nobody has used them in years. And sometimes nobody really knows anymore what the checkboxes on the selection screen actually do, now that the colleague who wrote the report has retired.
Unused reports, enhancements and modifications still have to be considered in security reviews, compliance requirements and system upgrades. Of course there are many advantages to staying as close to the SAP standard as possible.
But a system that is closer to the standard is not automatically a better system.
The next generation is already on its way
Greenfield and bluefield approaches can be exactly the right way to break up grown structures and realign processes. But even after a successful transformation, new requirements arise that are solved with custom developments. As soon as the ticket system opens, the next change requests come in — and with them the next generation of custom code.
For me, then, what matters is not how much custom code a system contains. What matters is whether we understand which functions are in it, what business purpose they serve and how they connect to the rest of the system.
Because the original purpose of a development is often lost while the code itself lives on. The technical what can often still be worked out. The business why is much harder to reconstruct.
Analysing code is not understanding a system
The possibilities for analysing and evolving SAP code with AI are also advancing fast. But analysing code is not the same as understanding an SAP system. An AI agent can have access to a development system and still not know which business function sits behind the code, which process it is used in, or why it exists at all.
That is exactly why we need not just better code-analysis tools, but an understandable knowledge base covering the entire system. Individual developments, processes and technical dependencies have to come together into one understandable picture — for people and for AI.
A central building block of Conjola is the feature map: it makes captured functions and custom developments visible and places them in their business context. The custom-code analysis uses AI to describe what each custom development does in business terms and groups them by topic. This knowledge helps the people who run, change and evolve SAP systems — and through the MCP access, your AI agent queries it directly, read-only and under your rules.