A director on our ARB asked a simple question last quarter: how many identity providers are we running in production, right now, today. Three architects in the room gave three different numbers, and none of them matched the CMDB export somebody pulled up on the spot to settle it. The meeting stalled for ten minutes while people argued over whose spreadsheet was current (mine wasn’t, for what it’s worth). That’s not a staffing problem, and it’s not a communication problem either. That’s what happens when the inventory of what you actually run lives in five places and none of them is authoritative.
The Fact Sheet Model Is the Whole Point
LeanIX isn’t a modeling tool in the ArchiMate-diagram sense, and I stopped expecting it to behave like one a long time ago. What it does well, better than anything else I’ve used, is the fact sheet model: applications, capabilities, tech components, and business processes wired together with real relationships, not metadata tags that go stale the week after someone types them in. Feed it from your CMDB, ITSM, and cloud billing integrations instead of asking humans to keep it current, and the inventory stays honest, because it’s pulling from systems that already have to be accurate for other reasons. That’s the difference between an EAM tool and a wiki page with a nicer skin.
The Waste You Can’t See Is the Waste You Pay For
The scale of the problem this solves is bigger than most leaders assume walking into the role, myself included. According to Zylo’s 2026 SaaS Management Index, the average organization runs 305 applications at a cost of $55M a year, with 46% of licenses unused, which works out to roughly $19.8M in annual waste, and IT departments controlling only 15% of that spend and 13% of those applications. You cannot rationalize what you can’t see. A live, relationship-aware portfolio view is the only way I’ve found to catch the sixth overlapping CRM, or the third identity provider, before it quietly renews for another year, which is the same discipline behind Reuse, Buy, Build: The Order Most Companies Get Backwards. The tool doesn’t make the rationalization call for you, and I wouldn’t want it to. It just removes the excuse that nobody could have seen the duplication coming.
The Bill You Don’t See Until Later
Here’s what doesn’t come up in the demo. Once your fact sheet taxonomy, your tagging conventions, and your integrations are built, you own that structure, full stop. Moving to a different EAM platform later isn’t an export job, it’s re-mapping hundreds of relationships by hand, and the sunk cost grows every quarter you use the thing well (which is a strange kind of trap to notice you’re standing in). Licensing scales with the size of the organization too, and if nobody is assigned to own data quality, you’ve bought yourself an expensive directory that architects quietly stop trusting. That’s the exact failure mode of an unmaintained CMDB, just with a better logo.
LeanIX Isn’t the Only Name on the List
There’s a vendor-direction risk layered on top of that, and it got sharper this year. At Sapphire 2026, SAP built its new AI Agent Hub directly on top of LeanIX as the governance layer for AI agents across the enterprise, not just applications, according to coverage of the announcement. That’s a genuinely useful extension if you want agent inventory and lifecycle governance out of the box instead of standing up a parallel tracker, and I’ll take it. It also means you’re increasingly betting on SAP’s product roadmap, not just an EA vendor’s, and those aren’t the same bet. LeanIX isn’t the only credible option on the table, either: Gartner’s 2025 Magic Quadrant for Enterprise Architecture Tools names Ardoq, Orbus Software, and Sparx Systems as Leaders alongside SAP. SAP’s recognition as a Customers’ Choice in the 2026 Gartner Peer Insights Voice of the Customer for Enterprise Architecture Tools is a reasonable data point for due diligence. It’s not a substitute for checking fit against what’s already generating your architecture data, and I’d tell any architect the same thing before they take my word for any of this.
Decisions Belong Next to What They Govern
LeanIX doesn’t ship a native ADR module, and at some point I just stopped waiting for one to show up. What changes is where the decision lives. Instead of an ADR sitting in a wiki disconnected from the component it governs, I attach it to the fact sheet of whatever it affects, so any architect reviewing that application sees the decision and the reasoning behind it right next to the thing itself, not three clicks and a search query away. The same goes for ARB exceptions: when a review board grants a time-boxed exception, the expiration date belongs on the fact sheet, not in a separate register nobody ever reopens, which is the exact discipline gap I called out in The Architecture Review Board: Narrow Docket, Small Room, Hard Exceptions. A decision that isn’t tied to the object it governs isn’t really captured. It’s just filed somewhere, and filed is not the same as findable.
The Next Object Type Is Already Here
The AI Agent Hub announcement is worth sitting with for a reason that has nothing to do with SAP’s roadmap. The logic behind building it on LeanIX instead of as a standalone product is that governing agents at scale is structurally the same problem as governing applications: discover it, map it to a capability and an owner, set a lifecycle state, retire it when it’s done. If your organization is currently tracking AI agents in a spreadsheet next to your application inventory, that’s the same failure mode restarting for a new object type, and I don’t think most people notice it happening until the sprawl and the orphaned ownership show up the same way untracked SaaS apps do today. Whatever EAM tool you’re committed to, the object model needs to extend to agents now, while the count is still small enough to govern on purpose.
I’m not neutral on this, and I don’t think architects who own the outcome should be, either. Pick one system of record, wire it to the systems that already generate accurate data, and require every ADR, every rationalization call, and every ARB exception to live inside it, no side channels. Switching later will cost me, and I’ve made peace with that trade, because the alternative is worse: architecture decisions that exist nowhere anyone can find them. For my organization, that system is LeanIX, and if I had to make the call again tomorrow, I’d make the same one.