Agentic AI Consulting

AI agents are transforming the way companies work with data, develop data platforms and operate existing systems. The key question here is not whether agentic AI is technically feasible, but where it creates tangible value and how it can be securely integrated into the existing data landscape.

We support you in identifying suitable use cases, developing the appropriate architecture and implementing initial scenarios in practice. In doing so, we consider not only individual AI models or tools, but also the interplay between the data platform, the agentic environment, MCP connectivity, semantic context, permissions & governance.

Whether it’s SAP BW, SAP Datasphere, Databricks, or a hybrid landscape with other data platforms, together we’ll develop an approach that fits your existing systems, processes and security requirements.

Identifying the Right Use Cases

Agentic AI in data and analytics has three very different perspectives. Business units want to “talk” to their data; controlling and planning want to calculate scenarios through dialogue; and IT teams want to develop and operate data platforms using agentic approaches. Each of these requires its own setup, different tools and different stakeholders. The first step, therefore, is to clarify which one matters to you, which tools in your environment are suitable for it and whether the necessary context is already in place or needs to be established.

We’ll help you choose the right approach, assess the tools and maturity level of each field, and use that to develop a realistic plan for getting started.

Bringing Systems, Agents & Context Together

A powerful language model alone is not enough. For an agent to work reliably, it needs access to the right systems and, above all, to the business and technical context of your data landscape.

In addition to the semantic layer of a data platform, this includes, for example, modeling rules, naming conventions, project knowledge, responsibilities and cross-system relationships. This is precisely where context engineering comes into play.

We help you build this knowledge base in a structured way and connect the agent to the relevant systems via appropriate interfaces such as MCP.

Security and Governance for Production Use

As soon as AI agents begin operating in production systems, every IT management team faces the same questions: Who is authorized to do what, who approved it, who can track it and what data is actually fed into a model?

An agent should only be able to perform actions that the logged-in user is authorized to perform. Read and write functions must be clearly separated, critical actions must be controlled and activities must be logged in a traceable manner.

Especially in existing SAP landscapes, this principle can be combined with existing authorizations and role-based access. The BW Modeling MCP Server uses the same interfaces as the BW Modeling Tools in Eclipse, so BW authorizations continue to apply with every call. Three additional roles on the BTP determine which tools the agent can even see; for example, a developer’s AI can be restricted to read-only access, even if the developer themselves is authorized to write in BW.

Strategy and Use Case Workshop

Together, we’ll analyze your existing data and analytics landscape, determine which perspectives matter most to you, and identify the areas where Agentic AI delivers tangible benefits—with an honest assessment of what’s possible today and what’s still on the roadmap. In doing so, with a focus on SAP BW, Datasphere, and BDC, as well as Databricks and hybrid data landscapes, we’ll examine the following topics:

Agentic Analytics for Business Units
Agentic Business Planning for Controlling and Planning
Agentic Development and Operation of Data Platforms for IT Teams

The result is not a general AI concept, but rather a prioritized selection of specific scenarios, an assessment of the appropriate tools, and a proposal for the next steps.

Proof of Concept and Technical Implementation

We don’t stop at concepts, we work with you to technically implement selected scenarios. To do this, we connect an AI agent to your existing data platform, for example, with SAP BW using the open-source BW Modeling MCP Server, or with Databricks using the official AI Dev Kit. This allows the agent to analyze your system, create objects, and execute tasks directly within the system.

In doing so, we make use of the existing interfaces and authorization concepts of the respective platform whenever possible, rather than setting up parallel access paths.

Context Engineering and Knowledge Base

An agent can only deliver results as good as the context it encounters. The semantic layer of your platform describes what the data means. We work with you to build everything that isn’t already in any system so that agents can use it on an ongoing basis.

This includes, in particular:

Capturing project knowledge, responsibilities, rules and naming conventions
Building a versioned knowledge base that can be used by both humans and agents
Rules for the agent and a way to indicate which knowledge is verified and which is not
Integration with the semantic layer of your platform and the systems that contain your context

This ensures that agents not only function technically correctly but also fit seamlessly into your environment.

Let's find the right starting point together

Would you like to find out what possibilities Agentic AI offers in your data landscape?

Whether it’s your first use case, a proof of concept, a production-ready MCP architecture, or a specific scenario in SAP BW: We’ll work with you to analyze your current situation and show you the best next steps.

Talk to our expert and let’s work together to discover how Agentic AI can support your data and analytics landscape.

From the initial idea to the productive deployment of AI agents

The Challenges