The automatic detection engine is good, but an inventive company name or an internal project name can slip past it. The vault’s dictionary fixes that: any term added to it is masked systematically, forever, in that vault.
Automatic detection versus dictionary
- An automatic detection is an inference by the local engine: excellent, but probabilistic. In the analysis panel it carries an “auto” badge and a dotted underline.
- A dictionary entry is a rule: deterministic, applied on every analysis, shown with a solid underline. Never a probability again.
Adding a term
Two paths:
- From the analysis panel (the most common): select the term in your text, right-click, choose its type (person, organization, place, project, and so on). Re-analysis is automatic and the term is masked immediately.
- From the Dictionary button in the vault header: browse the list and add terms directly, even before writing anything.
Tip: when opening a vault for a new client, immediately add the company name, key people and project names. Two minutes that secure all your future sends.
Removing or correcting a term
- Removing a mask from the dictionary is a more consequential gesture than dismissing an automatic suggestion: the app asks for an extra confirmation.
- To correct an entry’s type (a project classified as an organization, say): delete the entry and recreate it with the right type.
Good to know
- The dictionary belongs to its vault: “Client Alpha” terms never influence analyses in other vaults.
- After any dictionary change, an already-analyzed draft must be re-analyzed before sending: the app enforces this, so the approved preview always reflects the current rules.