AI Use Cases with Enterprise Architecture

1️⃣ 𝗗𝗲𝗺𝗮𝗻𝗱 𝗠𝗮𝗻𝗮𝗴𝗲𝗺𝗲𝗻𝘁

Every team submits requests for new tools, apps and integrations. Rarely, someone systematically checks whether this already exists. The result is duplicate developments, redundant licenses and wasted budget.

What makes using AI different from a standard form is the experience itself. Instead of filling out field by field in SharePoint, users interact with an AI that asks smart questions to get necessary information and guides them through the process. The result is either a quick reply that there already is a solution or a higher quality demand. And because AI can handle volume, you actually want more demands coming in, not fewer.

To make this work, you'd need a chat interface where the functionality runs (e.g. vibe coded). The AI needs to be trained to think like a demand manager, asking the right questions and knowing what information is needed (to populate the Fact Sheet). From there, the demand that lands in SAP LeanIX needs to feed into the actual demand process so the demand manager can pick it up and process it further. Think of this as the first quality gate before human review.

2️⃣ 𝗖𝗼𝘀𝘁 𝗜𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲

Some LeanIX workspaces contain financial context at license level, but that is rarely the full picture. Granular cost of ownership, the financial consequences of architecture decisions, what it actually costs when an application relies on certain IT components or interfaces — that data exists somewhere in the organization, spread across contracts, cloud billing and IT controlling tools, but almost never in LeanIX in a usable form.

Connecting LeanIX to an IT controlling tool like Apptio is an important foundation. But the integration alone is not enough, because raw cost data rarely maps cleanly to LeanIX artefacts. A cost center covers multiple applications, an application has costs from multiple sources, and contracts sit in PDFs that no integration reads. This is where AI creates real value: resolving ambiguous mappings, reading unstructured sources and enriching the data with market benchmarks where internal data is missing.

Getting there is a process, not a setup. You start by giving AI access to the relevant sources and working through the initial mapping together. Which applications correspond to which cost positions, which IT components carry shared costs, where are the gaps. That mapping improves iteratively over time. Once costs are properly attributed, AI can run the analyses that were never possible before: which applications are disproportionately expensive relative to their business value or lifecycle status, where redundancies exist with a concrete savings figure attached, and which decommission decisions have the clearest financial case. Every architecture decision can now be evaluated with real cost data behind it.

3️⃣ 𝗔𝘂𝘁𝗼𝗺𝗮𝘁𝗲𝗱 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 𝗗𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝗥𝗲𝗰𝗼𝗿𝗱𝘀

Architecture decisions happen every day, in Jira tickets, in meetings, in code reviews. Some of it gets documented, but rarely consistently or in a place where it stays accessible long term. When people leave the company, a lot of that context goes with them.

AI can close this gap systematically. Every Jira work item that touches the architecture is automatically identified, classified and saved as an ADR in LeanIX with context, timestamp and owner. Developers change nothing about how they work. What you gain is a living documentation layer that grows automatically as work gets done. Less time spent on documentation means more time for actual architecture work, and the institutional knowledge that usually walks out the door stays in the system.

To implement this you need LeanIX ARP as well as the Jira integration (or other). The key step is mapping your project hierarchy so that high-level initiatives in LeanIX are synchronized down to story level in Jira. From that point, when a Jira story is linked to an application being modernized, AI evaluates whether the change is architecture-relevant and writes the corresponding ADR entry directly into LeanIX, capturing what was changed, when and by whom.

4️⃣ 𝗣𝗿𝗼𝗮𝗰𝘁𝗶𝘃𝗲 𝗖𝗼𝗺𝗽𝗹𝗶𝗮𝗻𝗰𝗲

Most organizations have been through at least one audit, which means there are findings, requirements and frameworks sitting somewhere in documents. The audit report lives in a folder, the portfolio lives in LeanIX, and nothing connects them until the next audit forces the issue.

Combining both could change the situation fundamentally. AI continuously maps regulatory requirements against the current portfolio and flags gaps before they become audit findings. For a department head, this means walking into an audit with a clear picture of where you stand rather than finding out what the weaknesses are after the report. You can prioritize remediation based on actual risk, demonstrate to the auditor that gaps have already been identified and addressed, and build a track record of proactive governance rather than reactive cleanup. The cost of failing an audit, whether in fines, remediation effort or reputational damage, is sometimes higher than the cost of addressing issues early.

The AI needs to understand the regulatory framework in the context of your specific LeanIX meta model. That means working through the mapping together with the AI upfront: which requirements apply to which types of artefacts, which fields carry the relevant information, and what constitutes a gap. Once that foundation is in place, the scanning and reporting runs continuously.

5️⃣ 𝗔𝗿𝗰𝗵𝗶𝘁𝗲𝗰𝘁𝘂𝗿𝗲 𝗚𝘂𝗮𝗿𝗱𝗿𝗮𝗶𝗹 𝗖𝗵𝗲𝗰𝗸

In organizations with small EA teams, checking the existing landscape against architecture principles tends to fall to the bottom of the priority list, not because it is unimportant but because there is simply no capacity for it. The principles exist, but whether the actual portfolio follows them is rarely verified systematically.

AI can take on that verification continuously. It checks the existing landscape against the defined ruleset, identifies where applications or technology choices are out of alignment, and generates findings with explanations. This makes it possible to act on actual evidence rather than assumptions, prioritize remediation where the risk is highest, and have a clear answer when stakeholders ask whether the portfolio follows the agreed architecture direction. For the EA team, it turns a task that was never quite finished into something that runs in the background and surfaces issues as they arise.