The Configuration Agent helps administrators assess an existing account configuration and suggest new configuration by translating natural language requests into action suggestions.
The agent always returns action suggestions that must be manually reviewed and approved by an administrator before they take effect. Every executed action is audited under the approving administrator's identity, so the account's audit trail remains complete and accurate.
The Configuration Agent currently supports:
- Content Classification policies
- Content Safeguards policies
- AI Safeguards policies
- Document Labelling policies
Functional coverage is being actively expanded. If you would like to participate in early access for the broadened capability scope, please contact the Egnyte Products team.
Accessing the Configuration Agent
Select Configuration Agent from the dropdown list in the sidebar.
Additionally, the agent can be invoked from Secure & Govern via the Create With AI button under settings for supported capabilities.
How It Works
The Configuration Agent follows the same principle across every request:
- The administrator describes what they want in plain language.
- The agent interprets the request and returns one or more action suggestions.
- The administrator reviews the suggested actions and approves or rejects them.
- Approved actions are executed and recorded in the audit trail under the approving administrator.
Depending on the pricing plan, some capabilities may not be available. In those cases, the agent can suggest alternative approaches. For example, AI Safeguards policies can instead use criteria such as location or metadata and not Sensitive Content tags.
Use Cases
The following list outlines examples of use cases the agent currently covers:
1. Evaluating an Existing Setup
This use case is especially relevant for accounts that have already configured Secure & Govern where:
- Administrators do not have time to review those settings due to competing priorities
- The IT team has evolved over time and newer administrators lack the original situational context
- Administrators are not comfortable evaluating the settings on their own
The agent reviews the current configuration and summarizes it in plain language, helping the team understand what is in place before making any changes.
2. Configuring Policies (Find & Protect)
In a Find & Protect workflow, the agent assists with creating Content Classification policies to locate information, and then Content Safeguards and AI Safeguards policies to restrict access to it.
The agent can help build the following policy types:
-
Content Classification – Term Dictionary
-
Content Classification – Regex Dictionary
-
Content Classification – Custom Policy
3. Clarifying Questions and Fallback
If the agent does not have all the details required to prepare an action suggestion, it asks additional clarifying questions and refines its response based on the answers provided.
If an administrator asks the agent to configure a capability that is not yet supported, the agent responds with a link to the relevant help desk documentation.
4. Reverting an Action
After an administrator confirms the execution of an action, they can ask the agent to undo it. The agent returns a corresponding revert action for the administrator to confirm.
5. Shortcut to Egnyte Support
When the agent detects that an administrator is facing a problem, rather than simply requesting an action proposal or when human assistance is explicitly requested, it displays an additional message with a button to create a ticket for the Egnyte Support team.