SAP Indirect Access and AI Agents: How Digital Access Licensing Works

SAP indirect access became one of the most talked-about licensing issues in enterprise software after a UK court ruling in 2017. SAP responded with a new model, Digital Access, which licenses documents created in SAP rather than counting every person who touches the system indirectly.

Now AI agents are bringing the issue back. Copilots, chatbots and automated workflows can read and create data in SAP on behalf of large numbers of people, often without anyone logging in to SAP directly. This article explains how indirect access and Digital Access work, why AI agents raise new questions, and how organisations can manage the risk.

What SAP indirect access means

Indirect access describes situations where people or systems use SAP software through another application rather than logging in to SAP itself. A customer portal that creates orders in SAP, a CRM that reads stock levels, or a warehouse system that posts goods movements are all examples.

Under SAP’s traditional licensing, use was largely measured by named users. The question in indirect access disputes was whether the people using those other applications also needed SAP named user licences, even though they never opened SAP.

Because modern businesses connect SAP to many other systems, the answer to that question can have a large financial impact. That is why indirect access has been a central issue in SAP licensing for nearly a decade.

The Diageo ruling and why it still matters

The issue reached the courts in 2017. As a recent legal analysis on Mondaq explains, the UK High Court ruled in SAP v Diageo that thousands of Diageo customers and sales representatives using Salesforce-based apps, which exchanged data with SAP, were indirect users of SAP ERP. SAP claimed more than £54.5 million in additional licence fees, and the case settled before the damages phase.

Diageo was not the only case. InfoWorld reported at the time that SAP was also pursuing a dispute with Anheuser-Busch InBev over similar claims, raising concern among companies using third-party platforms to access SAP data.

The Mondaq analysis notes that the court’s reasoning was that use and access can reach through whatever sits in the middle. That principle is exactly why AI agents are now attracting attention.

How SAP Digital Access licensing works

In 2018, SAP introduced Digital Access as a way to license indirect use by documents rather than users. Computerworld reported that SAP offered existing ERP customers three options: do nothing, exchange named user licensing for document-based licensing through a contract addendum, or, for S/4HANA customers, convert to an entirely new contract.

Under Digital Access, SAP counts documents created in its system by indirect use. Computer Weeklydescribed how SAP identified a set of document types representing common business outcomes. There are nine: sales, invoice, purchase, service and maintenance, manufacturing, quality management, time management, financial and material documents.

The model is generally based on documents created rather than data read, which made indirect use easier to measure. However, analysts quoted by Computer Weekly warned that over-licensing could become a new risk if organisations bought more document capacity than they needed.

Why AI agents raise new SAP licensing questions

AI agents connect to SAP in ways that look very much like the integrations at the heart of the Diageo case, but at a much larger scale. The Mondaq analysis sets out four patterns that create uncertainty.

The first is conversational assistants with ERP connectors, where a licensed user asks a question and then shares the answer with many colleagues who have no SAP licence. The second is agentic workflows that create transactions without a human in the loop, such as matching invoices and posting documents. The third, more relevant to Oracle, is service accounts used by AI agents on behalf of many end customers. The fourth is retrieval pipelines, where SAP data is extracted into a separate store and used to answer questions.

The analysis notes that neither SAP nor Oracle has published clear AI-specific guidance answering these questions, and that existing contract language was written long before AI agents existed. That ambiguity is the core of the risk.

AI agents and Digital Access document counts

For customers on Digital Access, automated workflows raise a particular question. If an AI agent posts invoices, purchase orders or journal entries into SAP, each of those documents may count under the model. The Mondaq analysis points out that SAP’s published materials do not clearly exclude AI-generated documents from the count.

An automated process can create far more documents than a team of people working manually, especially if drafts, retries and reversals are posted along the way. That can change an organisation’s document volumes, and therefore its licensing position, quite quickly.

It is worth modelling the expected document volumes of any AI-driven process before it goes live, and designing workflows to avoid creating unnecessary documents.

SAP’s own AI and indirect access

SAP is also building AI into its own products. The Mondaq analysis notes that SAP offers Joule base capabilities at no additional cost, while pricing some premium AI capabilities separately, in some cases through consumption-based AI Units.

This creates a useful distinction. Where AI is the vendor’s own product, there is a reasonable argument that licensing it should include the rights needed for the outputs it produces. The analysis suggests that customers should make that case in writing and seek to have it reflected in the contract.

Third-party AI tools are different. If an external copilot or agent connects to SAP, the existing indirect access and Digital Access terms are likely to be the reference point, so clarity in the contract is even more important.

How to manage SAP indirect access risk from AI

Start by reviewing your current SAP agreements, particularly the definitions of user, access and indirect use, and whether you are on named user licensing, Digital Access or a mix. Where the contract is silent on AI agents, that is a point to address at your next renewal.

Next, document how each AI deployment connects to SAP. Record which users prompt it, what data it reads, what it writes, and which outputs create documents. That record will be invaluable if questions arise during an audit.

Finally, if you are buying new AI capabilities from SAP or a third party, ask for written confirmation of how they interact with your existing licence metrics before you buy. Our SAP licence optimisation services include analysis of non-SAP applications connected to SAP and the digital documents that may need to be licensed.

SAP named users, Digital Access or both

Most SAP customers are on one of three positions. Some still license indirect use through named users under their original contracts. Some have adopted Digital Access for document-based licensing. Others have a mix, with named users for direct access and Digital Access for certain integrations.

Each position responds differently to AI. Under named user licensing, the question is likely to be whether people benefiting from an AI agent’s access to SAP need licences. Under Digital Access, the question is more likely to be how many documents the agent creates.

Knowing which position you are in, and what your contract actually says, is the starting point for any assessment. It also helps when comparing the cost of different licensing options for a new AI deployment.

If you are unsure which model applies, check your most recent SAP order forms and any contract addenda. Many organisations changed their position during a renewal or an S/4HANA move without the change being widely understood across the business.

Common SAP indirect access mistakes

The first mistake is assuming that because an AI tool only reads data, there is no licensing question. Under named user licensing in particular, the way information is shared after it leaves SAP may still be relevant.

The second is switching on automated workflows without modelling document volumes. Under Digital Access, a process that creates many documents can change the licence position quickly.

The third is relying on verbal assurances. If an account team tells you an AI capability does not affect your licensing, ask for that in writing and, ideally, in the contract itself.

Questions to ask before deploying AI on SAP

Before an AI agent goes live against SAP, it helps to answer a few questions internally and with SAP. Which licensing model covers our indirect use today, named users, Digital Access or both? Will the agent create documents in SAP, and if so, which of the nine document types, and how many?

It is also worth asking SAP directly how a specific AI deployment would be treated under your contract, especially if it uses a third-party tool. If the answer depends on details that are not in the contract, ask for them to be added at your next renewal.

Recording these answers before deployment is far easier than reconstructing them during an audit.

SAP indirect access FAQs

What is SAP indirect access? It is the use of SAP software through another application, such as a CRM, portal or integration, rather than by logging in to SAP directly.

What was the Diageo case? In 2017, the UK High Court ruled that Diageo’s use of Salesforce-based apps that exchanged data with SAP amounted to indirect use of SAP ERP. SAP claimed more than £54.5 million, and the case settled before damages were decided.

What is SAP Digital Access? It is SAP’s document-based licensing model for indirect use, introduced in 2018, which counts documents created in SAP rather than indirect users.

How many document types does Digital Access use? Nine: sales, invoice, purchase, service and maintenance, manufacturing, quality management, time management, financial and material documents.

Do AI agents trigger SAP indirect access? There is no clear AI-specific guidance from SAP. Agents that read or create SAP data on behalf of users may raise indirect access or Digital Access questions, so they should be reviewed against your contract.

How can I reduce the risk? Review your contract definitions, document how each AI tool connects to SAP, model document volumes and seek written confirmation of licensing terms before buying new AI capabilities.

Conclusion

SAP indirect access has been a major licensing issue since the Diageo ruling, and Digital Access gave customers a clearer way to license it. AI agents are now creating new versions of the same questions, often at a much larger scale and without clear guidance from SAP.

The organisations best placed to manage this will be those that understand their current licensing model, document how AI connects to SAP, model the impact on document volumes and address AI explicitly in their next contract negotiation.

If you would like an independent review of your SAP indirect access position or AI plans, contact the 2Data team.

 

More on the Blog