Last updated: May 25, 2026
DruMan Solutions LLC builds practical AI and automation systems for business operations, marketing, sales, and reporting. This policy explains how we approach client data when evaluating or building AI-supported workflows.
Data Minimization
We request only the information needed to understand, build, test, or support the agreed workflow. When possible, we use sample, redacted, synthetic, or anonymized data before using real client data.
Access and Permissions
We prefer least-privilege access, temporary access, named user accounts, and secure sharing methods. Credentials should not be sent through ordinary email or website forms. Clients are responsible for approving access and removing access when it is no longer needed.
AI Tool Use
When AI tools are used, we evaluate the sensitivity of the data, the purpose of processing, and the controls required. We do not intentionally submit highly sensitive, regulated, or confidential client data to AI systems unless the client has approved the use case, tool, and safeguards.
Human Review
AI outputs should be reviewed by a qualified human before being used for client-facing communication, operational decisions, financial decisions, legal decisions, employment decisions, medical decisions, or other high-impact uses.
Testing and Launch
AI workflows should be tested with clear success criteria before launch. Testing may include output review, error handling, fallback paths, access checks, logging, rollback steps, and owner training.
Retention
Project data is retained only as reasonably needed for delivery, support, business records, and legal obligations. Clients may request deletion of project materials where feasible and legally permitted.
Client Responsibilities
Clients are responsible for identifying regulated data, contractual restrictions, confidential information, security requirements, internal policies, and required approvals before data is shared or workflows are launched.
Contact
Questions about AI data handling can be sent to support@druman.ai.
#