Can a BYOM AI model tell your FI ticket from your MM ticket? A BERT-based classifier trained on SAP support tickets sorts incoming issues into 9 modules. FI, MM, SD, HR, ABAP, BASIS, INTEGRATION, and more with 82%+ accuracy, and gets smarter every ti...
This paper presents an enterprise-grade architectural blueprint that modernizes customer interactions by embedding a natural-language conversational AI directly into custom web portals built on SAP landscapes. The solution is built around two princip...
Knowledge Graphs in SAP HANA Cloud, delivered via A2ATL;DR. Joule is the trusted conversational front door to SAP. To make it genuinely useful for a specific business, it needs your intelligence: your data model, your relationships, your semantics. ...
What if your AI coding assistant already knew every relevant product standard before you typed a single line?
What happens when you combine SAP's Relational Pretrained Transformer with a multi-agent LLM system and ask it to predict World Cup 2026 match scores? Surprisingly decent football analysis — and one very honest timeout.
What happens when you combine SAP's Relational Pretrained Transformer with a multi-agent LLM system and ask it to predict World Cup 2026 match scores? Surprisingly decent football analysis — and one very honest timeout.
Introduction: The Illusion of Intelligence For the past several years, the tech sector operated under a singular assumption: if you make autoregressive Large Language Models (LLMs) large enough and feed them enough compute, human-level artificial gen...
SAP Joule for Developers, ABAP AI capabilities have already generated strong interest across the SAP ecosystem, enabling customers to achieve remarkable results, as demonstrated by the success story from Bosch.Following our Sapphire announcements, de...
Move beyond the AI hype and learn how to implement, activate, and scale SAP AI in real business scenarios. Join SAP experts for four days of live sessions, demonstrations, hands-on learning, and technical deep dives.
Claude Generates So Much - Here's How I Actually Keep UpAI is producing an almost comical amount of output these days. In a single session, Claude can scaffold a feature, write the tests, refactor two files, explain the architecture, and suggest thre...
Hand Pi an LLM, a tool, and a Markdown file. Pi hands back a working agent. In this blog, we turn that idea into a real SAP S/4HANA analyst - ask it a question in plain English, watch it pick the right APIs, chain the calls, and answer. Five files. Z...
Custom Sales Order Creation Agent Using Joule StudioHi Everyone!In this blog, we will see how we can create a simple custom Joule Agent to create a sales order using Joule Studio. We will also trigger an email notification once the sales order creati...
Document Grounding Joule Custom Agent using SAP AI Core and Microsoft SharePoint - No AWS S3 Needed
If your RAG quality has plateaued, the bottleneck is almost never your vector DB or your reranker. It is almost always your chunker. This post walks the strategies that actually survive production: how each one works, when to use it, when not to, and...
Post 8 — AI Product Management seriesIn December 1978, United Airlines Flight 173 ran out of fuel while circling Portland, Oregon. The aircraft was mechanically sound. The crew knew the fuel state was critical. The captain, fixated on a landing gear ...
Post 9 — AI Product Management seriesIn 2021, a research team at the University of Michigan published an audit of the most widely deployed sepsis prediction algorithm in American hospitals. Epic Systems had marketed the Epic Sepsis Model with a claim...
The PRD worked when the hard part was building. With agents, building is the easy part, a prototype takes an afternoon. The hard part is deciding what the agent should do, where its authority stops, and how anyone knows when it's wrong. That work spl...
"After 30 days, we'll move the agent to full autonomy." That's a schedule, not a safety criterion. A clean record usually means the hard cases haven't come up yet, not that the agent can handle them. Why autonomy has to be earned against evidence, an...
Most AI coding agents compete by adding more features. Pi competes by removing them. Built around just four tools and a 200-word system prompt, Pi takes the opposite bet from Claude Code: keep the core small, push everything else into extensions. Her...
I’ve been hearing a lot lately about fine-tuning being “dead,” but the reality is much more interesting. In this SAP Community post, I share what fine-tuning actually is, where it still shines, and why it continues to matter for real-world enterprise...
Every sprint, we had the same problem. A refinement meeting would go well — real decisions made, edge cases surfaced, implementation approach agreed. Then the ticket would get updated with a fraction of it. By the time a developer picked it up three ...
First, the content of this blog was initially written in Chinese by me (not an AI), but later it was translated into English with the help of AI, and the AI also enhanced and structured the content.Second, the content reflects my own experiences and ...
What if business users could simply ask questions about SAP S/4HANA data and instantly receive interactive visualisations in return? I built a prompt-driven AI agent on top of SAP BusinessObjects that makes this possible transforming natural language...
Alert is generated in Alert Inbox in English by default,You want to generate alert with other language e.g. Japanese.You want to forward alert to Third-Party via SNMP adapter with other language e.g. Japanese.
How I Built an AI Agent That Generates SAP Analytical Artifacts (CDS VDM → Cube → Analytic Query) via OData — Part 1 of 2: Architecture and First ArtifactHow I stopped writing CDS Views by hand and started generating the whole S/4HANA analytical stac...
Introduction:It is possible to connect to a remote PostgreSQL database using Python and then use your application to access and manipulate the data wherever you are. Visual Studio supports writing Python code, and using libraries such as psycopg2 to ...
Personal wikis are powerful — but as they grow, maintaining them becomes a burden. Andrej Karpathy's LLM Wiki concept flips the script: instead of manually keeping sources categorized, refactoring articles, and updating conclusions, you supply theraw...
Introduction:Creating an employee table in a Cloud PostgreSQL database using Python helps in organizing and storing employee information in a structured way. By using libraries like psycopg2, Python can easily connect to the database and execute SQL ...
We’re proud to share that ABAP FS now has 290+ GitHub stars and 53K active installs - gaining 10K new active installs since v2 went live! And this doesn’t even include offline versions probably floating around inside enterprise environments!
Moving on from the two very basic custom Joule Skills created in the previous post — which could only create and update ServiceNow incidents with a very limited and hardcoded set of input values — this blog takes a much bigger leap toward a fully aut...
Enterprise AI becomes significantly more powerful when it can access organization-specific knowledge instead of relying only on generic model responses.With document grounding in SAP Joule, Joule can retrieve information from enterprise repositories ...
In this blog I go over the scenario of creating an agent that based on user queries interprets them, translates mails and content to business partner and material, and then creates a sales order in the system. This is part 1, in parts 2 and 3 I will ...
Step-by-step guide to deploying Claude Sonnet 4, Claude Haiku 4.5, and GPT-4o through SAP AI Core and the Generative AI Hub — with working Python code, model selection guidance for SAP-specific use cases, and a practical comparison table. Updated fol...
Drift is not one thing. An agent can drift in four independent directions simultaneously: model, data, supervision, and scope. None of the four announces itself. The combination can be more dangerous than any single vector. This post covers the four ...
An agent that is not monitored does not sit and wait. It continues to act, autonomously, at scale, until a supplier calls to ask about orders that never arrived. This post covers the six observation instruments every agentic product needs, why automa...
The QA mental model that has served you for twenty years breaks in exactly three places when software starts making decisions instead of executing code. Pass@K replaces binary pass/fail. Compound probability means component accuracy and system reliab...
Introduction:Installing PostgreSQL to use it on the server will assist you in storing and processing data used by your application safely.It will imply the installation of PostgreSQL, the creation of a database, and the configuration of server access...
Claude Code ships powerful primitives - skills, memory, subagents, MCP servers. But primitives alone do not make a production system. This post teaches the harness architecture - seven composable pillars that turn Claude Code into a predictable, cont...
Most agentic products ship with only the agent designed. The supervisory system, how humans monitor, intervene, audit, and recover from what the agent does, is an afterthought. This post covers the four runtime artifacts every agentic product require...
Explore how AI transforms processes for your business using Customer Projects as an example. From opportunity to delivery and closure, AI helps you make better decisions, automate routine tasks, and gain real-time insights at every stage. Every succe...
Business cases for agentic AI are systematically 40 to 60 percent too low. Not because people are careless. Because the cost structure is genuinely different, token pricing is not what you expect, and the governance stack nobody budgeted is the line ...
The most expensive mistake in agentic AI does not happen in production. It happens before a single line of engineering is written, when a team decides to build an agent for a problem that should have been solved with a clearer dashboard or a better r...
A practical guide to working effectively with Claude Code. Covers workflow principles, planning strategies, model selection, and the full tooling ecosystem.Who This Is ForDevelopers — from first-time Claude Code users to power users. Part 1 covers pr...
"Build a A Agentic Hyrid RAG system that combines knowledge graphs and vector search — withself-correcting agents on SAP BTP & A step-by-step guide to building Agentic Hybrid RAG with Python, HANA Cloud, and AI Core."
Seven mental model shifts every product manager needs before any agentic AI project begins. Not because they are advanced. Because they are prerequisites. From suggestion vs. action to the two channels most teams never build — the concepts that separ...
Agentic AI is not just smarter software. It is probabilistic, it acts autonomously, and it does not know when to stop and ask for help. This series maps the frameworks product managers need to design the system around the agent, not just the agent it...
In one of the most crucial sectors – Aviation, safety relies on identifying and mitigating risks through robust, multi-layered systems designed to counter cognitive biases and organizational failures. Modern safety management integrates human factors...
SAP Joule isn’t a magical ITSM plug that sits between systems wearing a badge. It’s more like an enthusiastic translator at a networking event who says, “Oh, you’ve got an incident in ServiceNow? I’ll just whisper that into the API for you.”
| User | Count |
|---|---|
| 11 | |
| 8 | |
| 6 | |
| 6 | |
| 6 | |
| 6 | |
| 5 | |
| 5 | |
| 5 | |
| 5 |
GTLC_Vamsi
:
My productivity is boosted with Joule work desktop...
| Subject | Likes |
|---|---|
| 10 | |
| 6 | |
| 6 | |
| 6 | |
| 5 | |
| 5 | |
| 4 | |
| 4 |
