Keep customer data out of the prompt
A configured response can block a request or mask illustrated identifiers before an employee sends it on a supported path.
Workforce AI Security
AI Governance for Financial Services
Workforce AI Security for customer, KYC and operational administration in financial services.
For financial-services security, risk, compliance and IT leaders responsible for employee AI use.
Customer service
PromptUser or AI agent
Draft a reply to Priya Shah (0491 570 110) about account AC-883021.
Redacted
Identifiers removed
The illustrated identifiers are masked so the request can be reviewed in a safer form before sending.
Protected prompt
Draft a reply to [REDACTED NAME] ([REDACTED PHONE]) about account [REDACTED ACCOUNT].
Synthetic example
Sensitive content was detected in this illustrated request.
Make policy operation inspectable
Teams can review recorded responses by tool, group and data category as part of their internal governance work.
Separate workforce controls from regulated decisions
Aona controls employee interactions; the organisation retains its own decisioning, review and assurance processes.
Workflows
Keep customer and KYC identifiers out of AI prompts.
Start with the everyday AI interactions your people already need to complete.
Synthetic example
Draft a response from an anonymised case
Draft a reply to [REDACTED NAME] ([REDACTED PHONE]) about account [REDACTED ACCOUNT].
ActivityRedacted
- Group
- Financial Services
- Data
- Priya Shah · 0491 570 110 · AC-883021
- Response
- Identifiers removed
Sensitive content was detected in this illustrated request.
Aona does not make customer, credit, investment, AML or KYC decisions, and it does not establish regulatory compliance. Confirm supported tool and endpoint behaviour with synthetic data before rollout. Choose backend hosting and prompt processing separately; confirm the supported configuration and data flow before rollout.
Deployment, on your terms
Your infrastructure. Your processing choice.
Choose where the Aona backend runs, then where Aona processes prompts.
Host the Aona backend
Customer cloud, on-premises or Aona-managed hosting.
Process prompts with Aona
On-device, in your infrastructure or on Aona-managed servers.
Deployment answers
Scope the rollout with the real control boundary.
Confirm the tools, surfaces and data paths that matter before expanding beyond a synthetic test.
View details- 01Discovery
- Endpoint coverage is required for observed workforce AI use. The catalog tracks 10,000+ tools; catalog presence is not an enforcement promise.
- 02Controls
- Configured prompt and file controls apply on supported tool, surface and action combinations. Confirm the exact workflow with a synthetic test.
- 03Deployment
- The browser extension supports Chrome, Edge, Firefox and Safari. The native endpoint app broadens scope, but provider and transport depth differs; IT deploys it on Windows and macOS with its existing tooling.
- 04Review
- Use recorded policy outcomes to support internal review. They are evidence for your process, not an automatic compliance, classification or risk conclusion.
Buyer questions
Answers for the people accountable for workforce AI use.
Illustrated with synthetic data; confirm your workflow before rollout.
Aona tracks more than 10,000 AI tools in its risk catalog, but that discovery breadth does not mean every tool has an available control. The response depends on the tool, surface, action and endpoint. Confirm your required workflows in the evaluation.
On a supported path, the employee receives a clear response before sending. Depending on the configured rule, they may see a block or a prompt to use a redacted version. A hard block does not offer a continue option.
Choose backend hosting separately from Aona prompt processing. Host the backend in your cloud, on your premises or on Aona-managed servers. Process prompts on the user device/on-edge, in your cloud or on-premises, or on Aona-managed servers. Confirm the supported configuration, retention, telemetry and integrations for your rollout; these choices do not change a third-party AI provider’s data handling.
Choose one specific administrative workflow, a group and a simple rule for financial-services teams. Use synthetic data to check the experience on the intended tool and endpoint, then expand only after reviewing the results.
Yes. They are different workflows with different information categories and owners. Start with one synthetic example for each, confirm the endpoint response, and document the review path before expanding.
No. Aona does not make KYC, AML, credit, investment or customer decisions. It is a control for workforce interactions with third-party AI and should sit alongside the organisation’s existing processes.
Resources
Continue the evaluation
010203
Read the financial-services AI governance guide
Questions Australian financial-services teams can take into an AI governance review.
Check supported coverage
Review the practical tool, endpoint and action questions for your evaluation.
Plan a deployment
Use the deployment guide to prepare a measured rollout.
Focused evaluation
Test a customer or KYC administration workflow.
Bring a synthetic example and the data categories your teams need to protect. We will focus on employee controls, review evidence and the endpoints that matter to your rollout.
Book a financial-services demoCustomer service
- Customer, KYC or operations workflow
- Synthetic request and configured rule
- Employee response and audit view
- Coverage, residency and rollout questions
Synthetic example