How can we help?

AI Employees can support business operations using configured instructions, relevant knowledge, and available operational capabilities. Their access should remain aligned with the business environment and responsibilities assigned to them. An AI Employee should not be treated as having unrestricted access to every resource simply because it can understand natural-language instructions. What determines an AI Employee's capabilities? Relevant factors may include: 1. The business or workspace associated with the AI Employee. 2. Its configured responsibilities. 3. Available business knowledge. 4. Supported integrations. 5. Permitted tools and operations. 6. Applicable authorization checks. 7. Collaboration configuration. 8. Any required human oversight. Can an AI Employee perform any action requested by a user? No such assumption should be made. A natural-language request does not automatically establish authorization. The operation must be supported by the relevant feature and permitted by the applicable access controls. Can an AI Employee access confidential business information? An AI Employee may use information that is relevant and authorized for its tasks. Organizations should carefully review which knowledge sources and integrations are made available. Can customer-facing AI Employees access internal resources? Customer-facing AI Employees should be configured with appropriate restrictions. Internal operational information should not be exposed merely because an external user asks for it. Can an AI Employee delegate work to another AI Employee? Polyosync supports AI collaboration capabilities. Delegated tasks must remain subject to the relevant collaboration configuration and authorization boundaries. Does AI collaboration automatically increase permissions? No. Delegation should not be treated as permission to bypass access restrictions. What if an AI Employee needs access to an external system? The relevant integration must be configured and authorized. The AI Employee's ability to use that integration depends on the supported operational permissions. What about prompt injection? Instructions embedded in untrusted content may attempt to influence AI behavior. Organizations should not rely solely on written instructions to protect sensitive resources. Technical authorization and tool restrictions are also important. Recommended practice Configure AI Employees for specific responsibilities, limit unnecessary information exposure, and review their available integrations and operational capabilities.