Isolated workspace
Each customer's data, buyer-facing context, authorization settings, transaction records, and operating reports are isolated by default.
Security, permissions, and AI boundaries
Mekyro's trust model is not unlimited AI action. It is AI running automatically within clear authorization, policy gates, and data isolation.
Permission boundary
The official site only describes entries. Real authorization is enforced by authenticated workspaces and backend permission rules; frontend menus are not security controls.
AI automation boundary
AI can prepare suggestions, run Skill workflows, organize Email, cite evidence, and move tasks; high-risk commitments require policy gates.
Sensitive data handling
Token, API keys, VM credentials, payment information, and private knowledge do not appear on the official site or public copy.
Human intervention boundary
Humans handle high-risk exceptions, authorization changes, and cooperation confirmation instead of turning daily operations back into manual queues.
What AI can do automatically
Organize leads, draft replies, update state, remind next steps, and compile knowledge.
Quotes, inventory, delivery commitments, payment, and refunds go through policy gates or human confirmation.
Read sensitive credentials without authorization, bypass permissions, fake payment results, or sign irreversible commitments for people.