Redact files before
they reach AI

To redact files before they reach an AI tool, you need redaction that runs inline at the moment of upload, not a separate manual step. Aona inspects each file an employee attaches to ChatGPT, Copilot, Claude, Gemini, or any other AI tool, detects sensitive content such as PII, financial data, and source code, and removes it automatically before the file leaves the device. For Word (DOCX) and Excel (XLSX) files, the redacted copy preserves the original layout, tables, and formatting, so the document stays usable. PDF redaction is currently in development.

Aona provides automatic inline redaction for Microsoft Word (DOCX) and Microsoft Excel (XLSX) files today. Sensitive content is removed in place while the document's layout, tables, headings, and number formatting are preserved. PDF redaction is in active development and is not yet generally available. We never claim a capability is shipped before it is, so PDF is clearly flagged as on the roadmap.

Yes. For DOCX and Excel files, Aona redacts sensitive content while keeping the rest of the document intact: tables, headings, cell formatting, styles, and structure all survive. The employee gets a clean, usable copy to share with the AI tool instead of a flattened image or a broken file. This layout-preserving approach is what separates inline redaction from crude file blocking.

Not yet. PDF redaction is in active development on the Aona roadmap. Today, automatic layout-preserving redaction covers Word (DOCX) and Excel (XLSX) files. For PDFs and other unsupported types, Aona can still apply your policy, for example by warning the employee or blocking the upload, so sensitive data is not exposed while PDF redaction is being built. Talk to us if PDF support timing matters for your rollout.

Aona's detection accuracy is high. The platform achieves 97 to 99 percent detection across common sensitive data types and an aggregate classification accuracy (ACD) of 92.9 percent. Detection covers personally identifiable information (PII), financial records, source code, and confidential business content, with custom classifiers available for organisation-specific data types.

Yes. Aona applies file redaction policy consistently across ChatGPT, Microsoft Copilot, Google Gemini, Claude, and thousands of other AI tools, whether they are sanctioned or shadow AI. Because Aona operates at the browser and endpoint layer rather than the network layer, it can inspect and redact files before they are uploaded, even for AI tools accessed over encrypted HTTPS connections.

Aona is SOC 2 Type II certified, with security, availability, and confidentiality controls independently attested. File inspection and redaction happen inline before the file leaves the device, so sensitive content is removed at the source rather than after it has reached a third-party AI tool. This supports GDPR, HIPAA, and internal data-handling obligations for regulated organisations.