Modern game development has become slow, expensive, and fragile. Teams are larger, timelines are longer, budgets are higher, and even highly produced games still frequently fail. I think one reason is that content production has not scaled well enough. The industry still relies heavily on bespoke hand-authored content, even in genres where structure, variation, and validation could be much more systematic. What's more, the authored content rarely appeals to everybody, which often results in commercial disaster. (See, e.g., Veilguard.)
I'm working on a procedural authoring system for party-based cRPGs. The idea is to treat RPG content as structured, validated data rather than loose text or one-off scripting. A procedurally generated location has encounters, hazards, NPCs, rewards, state changes, and a validation report. A generated quest has a hook, possible approaches, consequences, rewards, and failure states. Main campaign questlines, which can be modeled on the old "Adventure Path" system from Dungeon Magazine, are typically revealed in a stepwise way. The system automatically plays through and rejects content that is logically broken, unrewarding, or impossible to complete.
The broader goal is to build better authoring tools for RPGs: Tools that help small teams create large, reactive games very quickly, without giving up structure or design control.
So I'm starting a new company to push the boundaries of procedural generation and related technologies in games. No link to share yet. Maybe in a month or two. We do have playable games already, though.