- What it answers from
- Your own documents, including text read from scanned and photographed pages, indexed email, and files in connected Google Drive, Dropbox and OneDrive folders. Not the open internet.
- Request types
- A direct question, a summary, a translation, a comparison between two documents, or specific details pulled out — all asked in plain language.
- Citations
- Every answer cites the document it came from and the line inside it, so the response can be checked against the source rather than taken on trust.
- Access control
- Permissions are applied before retrieval, not after generation. A document stays private to whoever added it until deliberately shared, and a record this reader cannot open cannot contribute to their answer or appear in its sources.
- Document currency
- Documents carry a draft, active or superseded status, so an answer can say which version is in force instead of quoting whichever copy was found first.
- Languages
- Nepali and English throughout, including asking in one language about a document written in the other.
- Regulatory monitoring
- New circulars and notices from the regulator are collected automatically and summarised as they are published, so they are answerable the same day.
- Cost visibility
- AI usage is measured per organisation and itemised on your bill, rather than sitting inside a flat platform fee.
- Tenancy
- Each organisation’s records are stored fully separated from every other organisation’s.
- In progress
- Related policies and past decisions surfaced together with an answer, so precedent is not lost. Being built, not available yet.