Give every company agent a clear role
Define purpose, knowledge, users and review rules before extending an assistant’s permissions.
Editorial analysis and illustrative technical scenarios. They do not define the features or integrations included in a UTOMAI subscription. See Private AI On-prem pre-launch information.
AI-generated editorial illustration. Not a product photograph or a customer case study.
A useful agent starts with a purpose a colleague can explain: help consult technical manuals, draft from product information or answer questions about approved procedures. That purpose tells the team which knowledge to provide and how to review the output.
Configuration before autonomy
Choose the intended users, approved material and instructions. Include requests outside the role, so the team can see whether the assistant acknowledges a limit. Dedicated Cloud or on-premise changes the agreed processing environment; it does not remove the need to define who may use the agent.
External actions need a separate review
A document-based assistant is different from an agent that changes a customer record or sends a message. A custom extension should specify the system account, allowed operations, authorization and human review. Access should follow the defined task rather than a broad administrator role.
Document operating responsibilities
Agree who maintains the knowledge, reviews instructions and handles a problematic answer. Where action traces are part of an integration, define their purpose and retention without collecting unnecessary content. Monitoring infrastructure is not the same service as reviewing every generated answer.
UTOMAI’s configurable agents sit within the platform offer. Bespoke agent development and system actions are additional services when outside the agreed configuration. A short role description keeps that distinction clear for users, IT and the people providing support.
