This is driving me crazy. We’re currently running Camunda for our main BPM work, but we’ve also got separate subscriptions for different AI models because various teams needed different capabilities—OpenAI for one team, Claude for another, then we added Deepseek because someone needed it for a specific project.
Now we’re evaluating open source BPM options, and every cost analysis feels incomplete because we’re trying to factor in all these separate AI model subscriptions on top of the platform licensing. It’s a mess.
Finance wants a clean TCO comparison, and I can’t give them one because I don’t even know how to properly account for the AI model costs in the migration scenario. Do we include them? Do we assume we’ll consolidate? And if we consolidate, which models do we keep?
Someone told me that there are platforms that bundle multiple AI models under one subscription, which would simplify this entire calculation. But I’m not sure if that’s actually realistic or just another sales pitch.
Has anyone actually done this consolidation? How do you model the actual savings when you go from managing eight separate subscriptions to one unified platform? And does that consolidation actually make your business case stronger, or does it just move the problem around?
I’m mostly interested in the real math—what actually changed when you simplified your AI model costs?