Just like last year, the entire INNOV8iON team traveled to St. Leon-Rot for UI5con and reCAP. Two days filled with new technology, real-world case studies, hands-on sessions, interesting discussions, and, of course, plenty of time for socializing. Once again, our base was historic Heidelberg: a great place to continue our conversations, exchange ideas, and have a drink together after a day packed with content.
UI5con took place on July 14, 2026, followed by reCAP on July 15. Both events brought together the SAP community and the development teams behind UI5 and CAP to share the latest developments, best practices, and real-world experiences.
What stands out most to us? AI was no longer just a single, isolated topic on the agenda. It served as a common thread running through virtually everything: from user interaction and workflow automation to development, architecture, and security.
Furthermore, significant strides have been made in both UI5 and CAP to make it easier for existing and new applications to integrate with AI assistants. As a result, AI is becoming increasingly integrated with the applications, processes, and business context that organizations are already using today.
UI5con: Smarter Applications and a New User Experience
UI5con wasn’t just about new controls or technical improvements. The central question was, above all, how we can build better applications: smarter, faster, easier to maintain, and more user-friendly.
The sessions on agentic integrations with n8n, offline capabilities, MCP Apps, AI-assisted development, and TypedJSONModel were highly relevant to the challenges we face in our projects.
Complex workflows without unnecessary manual work
If you want to build a more complex application, the solution usually involves more than just a good interface. Behind that interface, there often needs to be a workflow that brings together systems, users, documents, and decisions.
Tools such as SAP Integration Suite, SAP Build Process Automation, and n8n offer various options for orchestrating such processes. AI can be specifically incorporated at points where interpretation is needed.
Consider, for example, a request in which a user must upload multiple documents. The application can specify which types of attachments are required and automatically verify the documents after they are uploaded. Was the correct document submitted? Does it contain the necessary information? And does it meet the specified requirements?
The user then receives immediate, targeted feedback. This prevents documents from being sent back and forth multiple times before the application is complete. It shortens the turnaround time, reduces repetitive work, and creates a much more pleasant user experience.
The power doesn’t lie in adding AI just because it’s possible. The power lies in a smart combination of process automation, integration, and AI—used exactly where it’s needed.
Working offline starts with the use case
Offline functionality was also a major focus of the program. Making an application available offline sounds simple, but in practice, there are various strategies, each with its own advantages and disadvantages.
Should it only be able to load the interface without an internet connection? Should a user also be able to view and edit data? How much data do you store locally? When do you synchronize changes? And what happens if locally edited data conflicts with a change in the backend system?
The right solution depends entirely on the situation. A field technician working at a location without a stable internet connection has different needs than an office worker who occasionally experiences a brief network outage.
That is why offline functionality should not be added as an afterthought. It must be incorporated into the architecture and user experience from the very beginning. By outlining the various strategies and considerations in advance, you can make an informed choice for each use case.
UI5 apps directly in an AI chat
One of the most notable developments was the integration of UI5 and MCP Apps. This allows an AI assistant to provide more than just a text-based response. An MCP tool can also offer a pre-developed, interactive user interface within an AI environment that supports MCP Apps.
During UI5con, examples were presented ranging from simple HTML to UI5 Web Components, UI Integration Cards, an interactive UI5 linter workflow, and applications involving SAP transport data.
That opens up some interesting possibilities. Suppose a user asks an AI assistant which requests, orders, or shipments still require attention. Instead of a long list in plain text, the assistant can display a recognizable UI5 view with tables, filters, statuses, and available actions.
It is important that this interface is not generated anew by the language model every time. The organization develops and manages the UI itself. This allows you to maintain control over the design, available data, interactions, and actions.
Users thus enjoy the convenience of a natural-language conversation, combined with the reliability and consistency of a pre-built SAP interface.
In this way, an AI assistant becomes not just a standalone chatbot alongside the SAP landscape, but a new gateway to existing information and processes.
Faster Development with AI and MCP Servers
AI doesn’t just change what the user sees. It also changes how developers build applications.
The UI5 MCP Server provides AI coding assistants with access to relevant UI5 knowledge and development tools. This enables an assistant to provide more targeted support for questions about a project, the application of UI5 patterns, and the development or customization of applications.
During the hands-on session on the MCP Server for Fiori, participants also used AI coding assistants to create and customize SAP Fiori Elements applications. The accompanying materials demonstrate how different MCP servers can be combined for AI-powered full-stack development.
This can save developers a lot of time. An assistant can help set up a project, find relevant documentation, analyze existing code, and generate suggestions.
However, the developer remains the one who understands the context, evaluates choices, and is responsible for quality, security, and maintainability. AI does not replace craftsmanship, but it does provide professionals with more powerful tools.
TypeScript lets your code speak for itself
Another practical topic was the use of TypeScript and TypedJSONModel.
TypeScript makes expectations regarding data structures and functions explicit in the code. As a result, the development environment can detect errors earlier, and developers receive better support when writing and modifying applications.
TypedJSONModel brings that type safety to UI5’s JSONModel. This allows for better validation of binding paths and values during development.
This makes an application easier to understand and maintain. Developers don’t have to rely as much on external documentation or implicit conventions to understand how a model is structured. To a large extent, the code itself explains what data is available and how it may be used.
reCAP: Greater Focus on Business Context, Security, and Reuse
A day later, reCAP was entirely focused on the SAP Cloud Application Programming Model.
During the keynote, we gained insight into the strategy behind CAP and how the framework is continuing to evolve. The session also focused on topics such as data replication, caching, AI-assisted development, Spec-Driven Development, security, declarative programming, and plug-ins.
Security is part of development
Security is always important, but it takes on even greater significance in the age of AI. AI makes it possible to produce code and expand applications more quickly. That speed must never come at the expense of control and security.
During the security workshop, participants took a hands-on approach to common risks from the OWASP Top 10. The goal was not only to identify vulnerabilities, but also to learn how to prevent and resolve them in CAP and Node.js applications.
That’s exactly how we view security—not as a one-time check right before going live, but as an integral part of design, development, testing, and management.
Especially when AI generates or modifies code, a developer must continue to understand what is happening. Faster development is only valuable if the final solution is also secure, reliable, and explainable.
Data Replication and Caching for Stable Applications
CAP is often used to consolidate data from various sources into a single local domain model. In doing so, you must consciously decide how the application communicates with those underlying sources.
Do you retrieve your data in real time for every request? Do you replicate certain data to a local environment? Do you use caching? Or do you opt for a combination of these approaches?
Every strategy has implications for timeliness, availability, performance, and complexity. Not every application requires continuous real-time data. At the same time, there are processes in which you absolutely do not want to work with outdated information.
During reCAP, the various strategies for data replication and caching were explored in greater depth.
By making a deliberate choice for each use case, you can reduce dependence on back-end systems while ensuring that users have access to the right information. This results in applications that are faster, more stable, and better able to withstand temporary disruptions.
AI-Assisted Development for CAP Applications
More and more AI tools are also becoming available for the development of CAP applications.
The MCP Server for CAP helps AI coding assistants better understand a CAP project. For example, the assistant can investigate which CDS services are available, how entities are related to one another, and which CAP documentation is relevant to a development question.
This is valuable for both new and existing applications.
For a new application, an assistant can help set up the data model, the services, and the initial implementation. For an existing application, AI can help you gain a faster understanding of its structure, dependencies, and the patterns used.
Because the assistant has access to specific project and framework context, its responses are more relevant than when the model has to rely solely on general knowledge.
Here, too, human oversight remains essential. A technically sound proposal is not automatically the best solution for the user or the organization.
From Standalone Prompts to Spec-Driven Development
To develop effectively with AI, you need to keep the business context closely aligned with your AI agents.
A prompt such as “Build an application for approving requests” does not provide enough information. Who is allowed to submit a request? What statuses are there? What exceptions apply? Who is authorized to approve requests? What happens if a request is rejected? And what requirements must the result meet?
Spec-Driven Development provides a way to document that context in a structured manner. Requirements, rules, decisions, constraints, and desired outcomes become part of a specification that both developers and AI agents can use.
At reCAP, this topic was the focus of the session “Beyond Vibe Coding: Engineering Deterministic CAP Applications with SDD.” Examples are also available that combine CAP, Fiori, and UI5 MCP servers with specifications and additional instructions for the AI agent.
This makes AI-assisted development more repeatable and verifiable. The result is less dependent on a single cleverly worded prompt and better aligned with actual business needs.
Describe what you want, don’t program how to do it
CAP’s philosophy remains powerful: define as much as possible what you want to achieve, rather than programming every technical detail of how to do it yourself.
CAP offers an increasing number of declarative options and reusable building blocks for this purpose. This allows developers to focus on the domain, the business logic, and user value, while the framework takes over a larger portion of the repetitive technical work.
Less custom code usually means less maintenance, fewer moving parts, and fewer places where errors can occur.
The range of plug-ins is also growing. There are extensions for various SAP BTP services and technologies, and the community is also developing reusable components.
In addition, you can create your own building blocks and plug-ins to apply functionality securely and consistently across multiple applications. Examples include shared validations, logging, notifications, or integration functionality.
When properly designed, these building blocks not only accelerate initial development; they also make it easier to roll out improvements and security updates centrally.
CAP and UI5 make both existing and new apps AI-ready
For us, the most important takeaway from both events is the combination of CAP and UI5.
UI5 handles the user experience. With MCP Apps, recognizable, interactive interfaces and actions can become part of a conversation with an AI assistant.
CAP provides the business layer: data, services, actions, and business logic. Using the MCP Protocol Adapter, a CAP service can be exposed as an MCP server. According to the current CAP documentation, a service can be annotated with ` @mcp` for this purpose. This functionality is currently available as a beta.
MCP then serves as the standardized interface between the application and the AI assistant.
This creates a powerful combination:
CAP provides the business context and functionality, UI5 ensures controlled visual interaction, and the AI assistant helps users achieve their desired results more quickly.
This isn’t just interesting for new applications. In fact, it allows existing CAP services and UI5 applications to be gradually integrated with AI assistants. This means organizations don’t have to rebuild their entire application landscape to get started with AI.
Existing data, services, and campaigns can be made accessible in a controlled manner. New applications can be designed from the outset to be usable through both a traditional interface and an AI assistant.
AI will therefore not become an isolated entity within the SAP landscape. It will become a new way to work together with the processes and applications that are already in place.
What We Bring to Our Projects
At INNOV8iON, we don’t just watch these developments from the sidelines. We research new technologies, build prototypes, share knowledge, and turn these possibilities into practical solutions.
We never start with the tool. Using Design Thinking, we first map out the user, the process, and the business context.
Where do people get stuck? Which tasks take up an unnecessary amount of time? What information is missing at the critical moment? Where can automation help? When does AI truly add value? And where does human insight remain indispensable?
Based on those questions, we work iteratively toward a solution. Sometimes that involves a smart workflow using SAP Integration Suite, SAP Build Process Automation, or n8n. Sometimes an offline UI5 app is needed for on-site employees. In other situations, a CAP plug-in makes it possible to deliver functionality more quickly and consistently.
When it comes to AI, we also look at the big picture. With UI5, we can offer users a familiar and interactive experience. With CAP, we unlock business context, data, and actions. With MCP, we connect those components to AI assistants.
In doing so, we always maintain control over security, authorizations, governance, maintainability, and user experience. Not everything that is technically possible automatically adds value.
We immediately apply what we’ve learned from UI5con and reCAP to our projects and internal experiments. We try out new tools, discuss the possibilities with each other, and challenge one another to look beyond the first solution that comes to mind.
This gives our clients access to up-to-date SAP expertise and provides our colleagues with ample opportunity to delve into technologies that are reshaping the industry.
And yes, that also includes a few fun days with the whole team in Heidelberg. Because the best ideas don’t just come from sitting in front of a laptop—they also come from good conversations with people who share the same enthusiasm for technology.
Want to innovate together?
Would you like to make smart use of the latest SAP technologies and integrate existing or new applications with AI assistants? Could you use some help turning an idea into a secure, user-friendly, and future-proof solution? If so, please contact us.
Are you especially interested in building solutions using CAP, UI5, the cloud, integrations, and AI as part of a close-knit team? Then stop by for a cup of coffee. We’d love to talk more about technology, your ambitions, and the impact we can make together.
