As a Product Manager for SAP Build, focused on Early Adopter and Product Enablement, SAP Sapphire is one of the most important moments in the year – not just to present what we have built, but to hear directly from the people building with it. This year, the SAP Build PM team was deeply embedded across both events: in the ASUG pre-conference, the keynotes, the sessions, and the Agent Lab. We asked our PM colleagues to do what product managers do best – observe closely, listen carefully, and capture what they saw. This post is the result of that: a collective picture of what Joule Studio's debut at Sapphire looked and felt like, from the people who were there.
SAP Sapphire 2026 was held across two cities – Orlando (May 11–13) and Madrid (May 19–21) – and the energy in both was unmistakable. Joule Studio made its public debut at Sapphire. Thousands of customers, partners, and developers experienced it hands-on: some for the first time, some building on what they had already seen in Florida. The conversation that started in the US continued in Europe with the same momentum and the same conclusion. Enterprise AI has stopped being a promise and started being something you can build in an afternoon.
Orlando: Learning by Building
SAP Sapphire Orlando set the tone for what this year's event would be about. Christian Klein opened the keynote with a question that reframes everything: will SAP even be a software company in the future? His answer – that SAP is becoming a business AI company – was not defensive. It was clarifying.
The core argument landed clearly: AI keeps stalling in enterprises not because the models are lacking, but because they are trained on the public internet, not on your business. They have no knowledge of your processes, your data, or your compliance framework. SAP Business AI Platform – and Joule Studio within it – is built around fixing exactly that.
Joule Studio was introduced as an intent-based, deeply contextualized build environment for agents, open to any third-party model with no lock-in and both low-code and pro-code paths – with agent development free and backed by a €100M+ ecosystem investment. Throughout the event, there was a noticeable shift away from isolated AI use cases toward AI-infused end-to-end experiences. Attendees were not looking for another AI experiment – they wanted proof that this works in production. Real-world customer journeys, with Fujitsu cited as a concrete reference multiple times, drove that conversation.
The keynote by Dr. Michael Ameling, Jonathan von Rüden, Jan Oberhauser, André Christ, and Maria Kondratyeva featured a live Joule Studio demonstration that became the defining moment of the event. The team demonstrated a 3-way invoice matching agent built end-to-end on stage. The scenario was immediately relatable: a company processing 50,000 invoices monthly, chronic matching exceptions, and approvals consistently landing with the wrong people.
Joule Studio received a natural language description of the problem. What followed was automated and complete: SAP LeanIX integration discovering the relevant SAP systems, SAP Signavio applying 50 years of SAP process expertise to understand matching strategies and routing rules, generated orchestration workflows for exception handling and escalation, and Agent Hub providing governance from the first deployment.
The result: Accounts Payable (AP) exception cycle time reduced from 72 hours to 90 seconds. Five manual touchpoints eliminated to zero.
One audience reaction said it all: “I couldn't tell if it was a real system or a guided demo.” That comment – part skepticism, part disbelief – repeated throughout the week. The bar for proving production-readiness is high. Joule Studio cleared it.
The Agent Lab: From Skepticism to Conviction
If the keynote introduced Joule Studio, the Agent Lab made it personal. Over three days, more than 1,000 unique customers and partners queued up and nobody left early. The numbers are striking: 3,366 AI solutions created, 1,640 agents built, across Finance, HR, Supply Chain, and beyond.
The experience followed a consistent pattern: skepticism at the start, hands-on building in the middle, and genuine surprise at the end. The barrier was never ideas – it was trust that the platform would actually work. Once they saw their own business problem translated into a working design, that trust arrived instantly.
As Samir Hamichi from the Agent Lab team observed: “People were looking to know the added value from SAP, what [was] the differentiator. When they walked through the experience, they understood the power of tools and strategy.”
What Makes Joule Studio Different
Every AI coding tool on the market can write code. What they cannot do is understand your business. Generic tools operate at the file and repository level – they have no knowledge of your procure-to-pay process, your enterprise architecture, or the SAP capabilities already available to you. Context is lost at every handoff.
Joule Studio changes the starting point entirely. The SAP Knowledge Graph, process models from SAP Signavio, enterprise architecture from SAP LeanIX, and SAP Domain Models all flow automatically into every development artifact from the first session. A business analyst describes a problem in plain language. Joule Studio generates a structured, SAP-context-aware Product Requirements Document – a shared source of truth for business stakeholders, developers, and architects, aligned from the first output rather than after weeks of back-and-forth. That PRD becomes a Technical Specification; the specification becomes production-grade, clean-core-aligned code, deployable to a fully managed runtime.
The insight heard repeatedly across the event from customers: “This brings business back into the loop.” And from one partner architect: “Context at scale – Signavio, LeanIX. That is powerful.”
One participant shared on social media what struck with him most: the end-to-end completeness. From describing the intent to generating a PRD, functional and technical specifications, unit tests, and a deployable solution – all without leaving the platform. What surprised him most was that the platform itself guides you through the process, asking clarifying questions like a colleague would rather than expecting you to have all the answers upfront. As he put it, you are not just building agents – you are building them with the help of agents.
You are not building an agent that connects to your business afterward. You are building one that is grounded in it from the first sentence you type.
The Governance Conversation Nobody Skips
One exchange at the Orlando Agent Lab deserves to be named directly. An attendee, after seeing what agents could do, said: “Now I can cut half my company.”
Maria Kondratyeva pushed back immediately: “That is not what agents are for. Agents eliminate soul-crushing manual work – data entry, chasing approvals, reconciling exceptions – so people can focus on work that requires judgment, creativity, and relationship-building. The work only humans can do.”
But the fact that the question was asked reflects how seriously people are taking this. The governance features in Joule Studio – Agent Hub, LeanIX integration, and human-in-the-loop approval gates – are not compliance checkboxes. They are what separates enterprise AI from unmanageable risk. The governance story needs to be as prominent as the capability story – and across both cities, the sessions that included it landed better every time.
Madrid: The Momentum Crosses the Atlantic
One week later, Madrid picked up exactly where Orlando left off. Dr. Michael Ameling, Jonathan von Rüden, Jan Oberhauser, André Christ, and Soufia Maherzi brought the same invoice matching demonstration to a European audience and got the same reaction. A fresh crowd, the same disbelief turning into belief.
The Madrid Agent Lab wrapped up with 765 customers and partners, 1,350 AI solutions created, and 770 agents built. Different city, same energy: long queues, nobody leaving, and that consistent moment of surprise when an attendee's business problem became a fully designed agent before their eyes.
As Friederike Marby recalled: “Agent Lab was intense in the best possible way. The sheer volume of feedback, the questions, the ideas flying around. We had customers who flat-out didn't believe it was real, almost looking for reasons to dismiss it. But what really stayed with me was watching them discover, on their own terms, that it absolutely was.”
Final Thoughts
What struck us most, putting this post together, was the consistency of what our colleagues reported across both cities – independent observations that told the same story. The skepticism at the keynote. The energy in the Agent Lab. The moment attendees stopped asking “can this really work?” and started asking “what do I build first?”
That shift – from doubt to conviction, not through a polished pitch but through direct, hands-on experience – is exactly what we set out to enable. Joule Studio's debut at Sapphire was not just a product launch. It was the beginning of a much larger conversation between SAP and the enterprise community about what it means to build with AI in a way that is grounded, governed, and genuinely useful.
We are grateful to our PM colleagues whose impressions and observations brought this post to life – and to every customer, partner, and developer who showed up, asked hard questions, and built something.
What is Next
Across two cities and six days of hands-on building, the combined numbers tell their own story: thousands of agents created, thousands of attendees who arrived skeptical and left with a deployment plan.
Joule Studio is planned to reach General Availability in Q3 2026. The trial is free, and the starting point is a business problem you already know you have.
Want to watch the sessions?
– Listen to the keynote – The Beginning of Better
– Discover the Demo – The Business AI Platform for your future
– Watch the Deep Dive – From intent to execution: Build AI agents with Joule Studio
– Hear the Strategy – The enterprise path to agentic AI
All Sapphire recordings are available here
Register for upcoming webinars here
Ready to start building?
Visit Joule Studio to learn more and register your interest.
Explore the Autonomous Enterprise Learning Journey for guided tutorials and deeper dives into specific capabilities.
The enterprise AI era is not approaching. In Orlando and Madrid, SAP customers and partners are already building it, one agent at a time.
