How can we help?

AI Employees may use connected business capabilities when those capabilities are supported by their configuration and the relevant authorization controls. However, connecting an external application to a workspace should not be interpreted as granting unrestricted access to every AI Employee. The ability to use an integration depends on the available platform functionality, integration configuration, and applicable permissions. Why does integration access matter for AI Employees? External integrations can provide access to business information and operational capabilities. Depending on the integration, an authorized operation might retrieve information, create a record, update an existing resource, or perform another business action. These operations may have real consequences for the organization. Does every AI Employee automatically receive access to all integrations? No such assumption should be made. Access must be evaluated according to the relevant configuration and supported authorization mechanisms. Can AI Employees access information from external services? Where supported and appropriately configured, connected information may be available to an AI-powered operation. The specific resources and actions depend on the integration. Can an AI Employee modify external business records? Only where the relevant integration supports the operation and the required authorization is available. Organizations should evaluate the operational impact of write actions separately from read-only access. What is the difference between read and write access? Read access allows information to be retrieved. Write access allows information to be created, changed, or removed, depending on the provider's API. Write access generally introduces additional operational risk. How should organizations approach AI Employee integration access? Step 1: Define the AI Employee's business responsibility. Identify the specific task or process the AI Employee is expected to support. Step 2: Identify the required external information. Determine whether the task requires access to an external service. Step 3: Determine the required operations. Separate information retrieval from actions that modify external systems. Step 4: Review the integration's available permissions. Confirm what the external provider allows. Step 5: Verify supported Polyosync controls. Confirm how the relevant operation is authorized and exposed to the AI Employee. Step 6: Test the intended behavior. Use controlled scenarios before introducing important production operations. Step 7: Review the results. Confirm that the operation produces the expected business outcome. Step 8: Reassess permissions when responsibilities change. Remove unnecessary access and review newly required capabilities. Should sensitive operations require additional oversight? Organizations should consider human review or approval for actions that involve sensitive information, financial consequences, irreversible changes, or other significant business risks. Whether an approval mechanism can be enforced automatically depends on the specific workflow implementation. Can an AI Employee bypass external provider permissions? An AI Employee should not be treated as having greater authorization than the credentials and integration mechanisms used for the operation. External provider restrictions remain relevant. Recommended approach Grant access based on clearly defined business responsibilities. Begin with limited, low-risk operations and expand only after validating the intended behavior and authorization boundaries.