About
Experiment for entering arbitrary text and retrieving it as structured data (and not with vector embeddings / similarity search). The data is stored and retrieved as document data with dynamic, context-based schemas on store and retrieval. The project will be implemented with a few "toy" ideas, like a workout tracker, shopping list, etc. to see how good it is at creating schemas based on the context, and whatever else is stored in the db (e.g. multiple workouts should ideally have the same schema, and it'll change if the data gets more complex). Eventually the work is to support a microbiology wet lab inventory system that needs to support things like bacterial/viral data, mutations/engineering/drift, arbitrary new columns we'd like to add over time, and general inventory stuff like reagents, software/versions, etc. If it works, then ideally it can be used as a generalized CRM/CMs for any kind of data, with a completely malleable schema.
