Per-dataset interpretation rules the AI uses on every chat answer. Encode revenue formulas, pricing targets, labor rates, company scopes — anything you want the assistant to remember as ground truth. Open a dataset to view, add, edit or deactivate entries.
How this gets used
Every active entry for a dataset is prepended to the AI's system prompt when answering questions about that dataset. Use it to encode interpretation rules that should override the model's default behavior — pricing targets, custom revenue definitions, company scoping, anything tenant-specific.