Where is our data stored?
+
In the region you choose — Hong Kong or Singapore — and never replicated outside it. Retention and deletion windows are configured to your policy rather than ours, and deletion is verifiable: you can ask us to prove a record is gone and we can show you.
Is our client data used to train AI models?
+
No. Never used to train or fine-tune a model, and never pooled with another firm's book. Your data is read to prepare your day and stays inside your tenancy. This is a structural property of how Mercury is deployed, not a setting someone could switch off by accident.
How is data isolated between firms?
+
Each firm runs in its own single-tenant environment. No shared databases, no pooled storage. Encrypted with TLS 1.3 in transit and AES-256 at rest, with customer-managed keys where your policy requires them.
Can the AI act without human approval?
+
No. Rebalancing, trade preparation and settlement run end to end, but every regulated decision point stops for four-eyes review and a named human approval. Software decides what deserves attention; a person decides what happens. Nothing reaches a client without that signature.
How do we produce evidence for an audit?
+
Call reports, approvals and reviews are captured when the work happens rather than reconstructed afterwards, each bound to the household, proposal version and approver it belongs to. Filter to the date range a questionnaire asks about and export it. Records can be read and exported, never edited or deleted.
Which regulatory expectations is Mercury built to?
+
Mercury is built to align with the data-handling and outsourcing expectations set out in HKMA and MAS guidance, and with SFC requirements on record keeping and electronic data. One clarification worth making plainly: Mercury is a software provider, not a licensed financial institution. Your firm remains the regulated entity. Mercury is designed to make your existing obligations easier to evidence — never to assume them on your behalf.