Where can we host Aona's backend?

In your cloud, on your premises or on Aona-managed servers. Choose backend hosting separately from where Aona processes prompts, then agree operating responsibilities, storage, retention and supported integrations.

Can Aona process prompts on the user device?

Yes. Aona prompt-processing options are the user device/on-edge, customer cloud or on-premises, and Aona-managed servers. Confirm the supported configuration, controls, file paths and telemetry for your rollout; the options do not imply identical capabilities or that all existing clients already use on-edge processing. The destination AI provider's data handling remains separate.

How do we enforce an AI acceptable-use policy?

Translate the relevant requirement into configured controls for a defined tool, data condition, employee group and supported action. Test those controls and the endpoint response. Aona does not automatically convert an acceptable-use document into an enforced policy.

What is the difference between Block, Redact, Warn and Flag?

Block stops a supported interaction; Redact changes sensitive content on supported paths; Warn is a nonblocking employee warning; Flag records evidence without the same employee intervention. Availability depends on the deployed endpoint and workflow.

Does a successful Playground test prove endpoint enforcement?

No. It helps examine the configured rule’s response to a test input. You should also verify the browser or native-app action, the employee notice and the recorded result on the specific release and workflow you plan to deploy.

Can we apply different rules to different teams?

Aona supports policies for tools and teams. The demo will confirm the group assignment and supported employee workflow for your deployment.

What should employees see when an AI prompt is blocked?

The shared notice explains that the prompt cannot be sent, shows the detected information and attributes the intervention to company policy. A hard block does not offer a continue override. Validate the message and supported endpoint behavior with a synthetic prompt.

Does enforcement cover every AI tool?

Coverage depends on the tool, employee surface and action. Review the coverage guide and confirm your required workflows in the demo.

Can we test a sensitive-data rule?

Start with a synthetic example of the information you need to protect and the outcome you want to assess. We will confirm the supported testing path for that use case.

Solutions / Policy enforcement

On this page

For IT and security leaders responsible for employee AI use.

AI policy enforcement for employee workflows.

Configure sensitive-data rules for supported AI tools and teams. See the employee response. Review the outcome.

Book a policy demo

Review a sensitive-data rule, the employee response and deployment fit.

Try the policy example

FRAMEWORKS → GUARDRAIL CONFIGURATION

One rule. A clear employee response.

REDACTED

We protected your data

Sensitive identifiers are removed in this example. Review the illustrated changes below.

Redacted before sending

Person names: 1
Email addresses: 1

Redacted due to company policy

The plugin and app share this notice design. Coverage still depends on the deployed endpoint and action.

Basic Information

Guardrail NameCustomer identifiers

Description

Protect example names and email addresses in employee AI prompts.

SYNTHETIC PROMPT

Summarise the notes for Taylor Example. Contact taylor@example.invalid.

Illustrated response: identifiers redacted

Reconstructed from Aona’s shared enforcement-notice component with invented data. No prompt is sent, no file is processed and no device policy is changed.

Scope to confirmBrowser and native coverage varies by tool, operating system and action.

Coverage and deployment details

Controls depend on the AI tool, employee surface and action. Discovery coverage does not mean every tool supports enforcement.

Rules must be configured and deployed. A policy document is not automatically converted into controls.

Validate the required actions and file-handling path with synthetic test data before expanding the rollout.

Built for your decision

Start with the employee workflow.

Sensitive prompts, team rules or endpoint rollout: choose the decision you need to make.

Security leaders

Protect sensitive information in employee AI prompts

An employee includes a sensitive customer identifier in an AI prompt.

Synthetic example

AI prompt · test input

Summarise the renewal for Demo customer ID 48291.

Invented input

Conditional responseRedact

Identifier removed before this supported prompt path is sent.

Keep in viewUse invented data and verify the specific endpoint workflow.

Scenario details

Inspect

Data condition
Guardrail response
Employee notice

Next decision

Choose the supported action that matches your policy requirement.

Define the information that should trigger a response, such as a sensitive customer identifier, and confirm the supported prompt path. The employee notice should make the next step understandable. Test the rule with invented examples before using it to assess a real workflow.

From policy to logic

How to enforce an AI acceptable-use policy.

A written rule is the starting point. Connect the data, employee group and supported response in one testable example.

Choose a policy example

Illustrative logic · not a saved configuration

The rule to test

Sensitive customer identifiers

When: Customer email address
For: Finance
In: Supported browser prompt

Then, if supportedRedactRemove a customer email address before the example prompt is sent.

Synthetic prompt: [identifier redacted]

Validation checks

Illustrated browser prompt · Finance group

Test: synthetic identifier triggers the configured rule
Test: review copy contains no source identifier

Policy walkthrough

Reduced motion · choose a step

Confirm the endpoint and action before a pilot. Block, Redact and Warn depend on deployed support; Flag is audit-only.

Bring one acceptable-use requirement and the employee workflow it needs to protect.

Use the demo checklist

How it works in Aona

From policy to a clear employee response.

Global admin · FrameworksIllustrative example

FrameworksGuardrail

Assigned ruleSensitive identifier

Finance · selected supported tool

Employee message configured

Configure the action and employee message.

Workflow details

01 / Global admin · Frameworks

Configure the guardrail

Review the guardrail’s name and description, then configure its action and message in its policy context. Block, Redact, Warn and Flag express different outcomes; select the supported response that matches the requirement.

02 / Browser plugin / desktop app

Check the employee experience

Validate the notice on the specific supported interaction. A hard block has no continue override. A warning is nonblocking; a flag records evidence without presenting the same intervention.

03 / Playground + Activity & Audits

Validate before expanding

Use synthetic inputs to inspect policy behaviour, then verify the endpoint action separately. Review the recorded outcome and assignment before rolling the rule out to additional users.

Make the demo useful

An AI policy enforcement evaluation checklist

Bring an acceptable-use requirement, one AI tool and a representative employee workflow. Keep prompt enforcement and file handling as distinct tests.

Bring to the conversation

01One written requirement

02One employee workflow

03The endpoint owner

Policy intentWhich exact interaction should trigger the rule?

What to inspect

The data condition, employee group and AI tool.

Configured actionWhat should the employee be able to do next?

What to inspect

Block, Redact, Warn or Flag and the employee message.

Endpoint supportIs this supported in the release we will deploy?

What to inspect

Browser or native app, operating system and action path.

Outcome evidenceCan we demonstrate both the rule and its endpoint behaviour?

What to inspect

The synthetic test input, response and recorded policy outcome.

Review a sensitive-data rule, the employee response and deployment fit.

Book a policy demo

A closer look

AI policy enforcement questions, answered.

Your use case. Your demo.

See how your AI policy would work.

Bring one AI tool and a rule your team needs to apply. We’ll focus the demo on the control, the employee experience and your deployment requirements.

Book a policy demo

What we’ll cover

Your AI tool and sensitive-data rule
Policy configuration and team assignment
Employee response and policy outcome
Supported deployment and next steps

Workforce AI security

AI policy enforcement for employee workflows.

One rule. A clear employee response.

We protected your data

Start with the employee workflow.

Protect sensitive information in employee AI prompts

Apply AI usage policies to the right teams

Plan browser and desktop AI policy enforcement

How to enforce an AI acceptable-use policy.

Choose a policy example

Sensitive customer identifiers

Finance review boundary

Review before external sharing

From policy to a clear employee response.

Configure the guardrail

Check the employee experience

Validate before expanding

An AI policy enforcement evaluation checklist

AI policy enforcement questions, answered.

See how your AI policy would work.