Digital Worker Strategy — privacy and data handling

Publisher: Agentron. This notice describes Digital Worker Strategy 0.54.9. Your organization's EmploidAI agreement, configured model providers, and connected services determine the additional terms applicable to your deployment.

Information processed

The worker processes messages and attachments addressed to it, workspace files, selected knowledge documents, tool arguments and results, and task state. When the corresponding sources are configured, it also processes email threads and meeting archives delivered to its workspace. Extracted notes can contain people, organizations, projects, topics, decisions, commitments, and source references.

Where processing happens

The runtime and persistent workspace/state live on infrastructure selected by the EmploidAI operator. Model-driven tasks send relevant prompts, history, knowledge, and tool results to the model provider configured for that employee. Background extraction and Brain reflection may make additional model calls. Tools can transmit their inputs to the services they connect to. A local worker container does not imply that all model or integration processing remains local.

The plugin does not itself connect an arbitrary mailbox or meeting account. Access comes from the integrations, credentials, and permissions configured by the administrator. Marketplace distributes the plugin package; installing it does not by itself upload the worker's conversation history or knowledge to the Marketplace listing.

Storage and retention

Conversations, knowledge notes and graphs, artifacts, source-processing state, background job results, and Brain reports may be retained in the worker's persistent storage or the platform services used by the deployment. This plugin does not impose a universal retention period. Administrators manage workspace/state volumes, platform records, backups, and deletion under their deployment policies. Uninstalling a plugin is not a guarantee that persistent volumes, backups, or records held by a model or external service have been erased.

Review and control

Administrators can restrict tools and integrations and configure the employee's model, knowledge sources, and background work. Knowledge notes, source references, traces, and Brain proposals support inspection and correction. Model-generated extractions and inferences may be inaccurate; review them before using them for consequential decisions.

Contact EmploidAI support for deployment assistance or to route a question to Agentron. For access, correction, retention, or deletion requests involving your organization's data, contact your EmploidAI administrator first. Avoid including secrets or unnecessary personal information in support requests.