I have been using the most powerful coding models for ABAP coding for quite some time now. It's painful, but I persevere because this is how I learn the shortcomings of the models to build a coding agent that actually works.
Here's what months of research have taught me, in just a few sentences.
We spent months studying the patterns in mistakes LLMs make while writing ABAP code. While many teams frame the risks in terms of "hallucination", we've learned that hallucination is not the core issue in ABAP.
The real problems stem from how LLMs generalize programming patterns and how ABAP fundamentally differs from mainstream programming languages.
This blog breaks down:
If you're building CAP or RAP applications, you've probably asked ChatGPT to find APIs in the SAP Business Accelerator Hub.
I tested ChatGPT with a straightforward request and found three distinct failure patterns that make general-purpose LLMs problematic for SAP API discovery.
Consultants struggle to find standard SAP capabilities before developing custom code for several interconnected reasons:
We launched on Reddit and it crashed our app. ðŸ«
Claude Opus 4.5 was released yesterday and is claimed to be "the best model in the world for coding," so I was curious how it performs in ABAP development.
So I asked both Opus 4.5 (the newest model) and Sonnet 4.5 (the previous model by Anthropic) to create an ALV report for sales order analysis using traditional ABAP. See the complete prompt in the attached document.
My first impressions:
Yesterday, we opened up free access to our SAP Change Request agent for everyone who requested it earlier.
The response has been amazing! So we're expanding access limits and squashing a few bugs we found along the way.
Excited to share an early preview of ChromaSQL, a new SQL-like language that helps ABAP developers access SAP’s tribal knowledge in a few seconds instead of wading through SE11/SE84 and stacks of notes. 🎉