Mail goes out under your brand, so the controls matter more than the features. These are the ones that are built in, and the one position we state openly.
Human approval is part of the platform's foundation, not a setting. A message can only be sent after an account manager has recorded an approval, and that approval is stored with the approver's name and the material they assessed.
Every discount traces to a contractual rate or a configured band, or no discount is offered. Prices in replies and offers come from a calculation engine, not from a language model. If a price cannot be calculated, the message is not sent.
A suspected opt-out suppresses the contact before anyone reviews it, and hardens into a confirmed opt-out if nobody clears it within 14 days. If the suppression check cannot run, the send is refused rather than risked.
When a message is translated, every number must match the original digit for digit and glossary terms stay verbatim. Any difference forces a human review.
The scoring that decides which accounts to work is separated from the text that reaches customers, and a quality check scans every draft for leaks.
Each customer runs on its own deployment with its own database. A deployment cannot address another customer's data.
The current release is documented to due-diligence depth, including a register of known limitations with impact and owner. We share the limitations register, the subprocessor list and the feature catalogue with prospects during evaluation.
If a control you need is not on this page, ask. We would rather say "not yet" than let you find out later.
Security and data-protection questions: contact@strives.ai
Send an anonymised ERP export and get a one-page diagnostic in five working days: how many accounts are silent, what that is worth, and where a pilot would start.
