Every reply starts from what your brand already said.
Nuncio reads your support inbox, drafts in your brand's own voice from your own sent history, and never states a fact it can't verify against your knowledge base and your order data.
Six steps, every email, in order.
Nothing here is generic. Every gate exists because it caught a real problem in a real inbox, with real money behind it.
Voice and facts, never blended.
One model that both writes like you and states facts is a model that can write something wrong in a tone that sounds right. Nuncio keeps the two separate on purpose, and drafts only where they agree.
It knows when it doesn't know.
Confidence is computed, not assumed, from three things Nuncio can point to. That's what makes letting a reply go out unattended defensible instead of reckless.
Auto-send is configured per category, per brand. Missing data, a stale order snapshot, an unrecognized sender, or no explicit permission all fail closed, every time.
The system doesn't stay still.
An edited draft becomes the new answer. A gap in the knowledge base becomes a proposal. Nothing here needs a retrain, and nothing seeded can be overwritten by the system.
The inbox is a dataset, not just a queue.
Nuncio clusters inquiries into trends, so a pattern is visible before it's a meeting.
Illustrative — the same shape Nuncio surfaces in a weekly digest.
Point Nuncio at the inbox you already run.
Bring us the mailbox. We'll show you what voice it captured and where the knowledge base has gaps, before you commit to anything.