This is an interesting case of a RAG application and will create more of a moat than it might first appear.
Since standard LLM training data doesn't know how to read binary game save files or how to correlate their contents into narrative content or suggestions, a whole separate training will be needed for every game that wants to support this.
Essentially, this service would turn MS into the publisher of on-demand walkthroughs and game guides, since the database will need to contain all the info those do now.
This is an interesting case of a RAG application and will create more of a moat than it might first appear.
Since standard LLM training data doesn't know how to read binary game save files or how to correlate their contents into narrative content or suggestions, a whole separate training will be needed for every game that wants to support this.
Essentially, this service would turn MS into the publisher of on-demand walkthroughs and game guides, since the database will need to contain all the info those do now.