Sentinel

Fraudulent support claims backed by AI-generated photos are on the rise. Fake product damage, fabricated receipts, and synthetic proof-of-purchase images slip past manual review — costing businesses thousands in false refunds.

Sentinel adds an automated layer of protection directly inside your Zendesk agent workspace. Every image attached to a ticket is analyzed by a multi-engine AI detection pipeline, giving your agents instant, actionable insight into whether an image is authentic or synthetic.

Key features:

• Multi-engine detection pipeline — combines complementary AI analysis engines for higher accuracy than any single detector alone.

• Automatic background scanning — new ticket images are scanned as they arrive. No manual work required.

• Inline and attachment support — catches AI-generated images whether uploaded as attachments or pasted inline into comments.

• Risk scores and indicators — each image receives a 0–100% AI probability score with human-readable indicators explaining why it was flagged.

• Internal notes — optionally post automatic internal notes on tickets when AI-generated content is detected, so the whole team stays informed.

• Team accounts — invite multiple agents to a single Sentinel organization so your entire team shares one subscription.

• Manual scan on demand — agents can trigger a scan on any image with one click.

Built for customer support, trust & safety, e-commerce, and insurance teams that handle photo evidence in tickets. Try free with 5 scans — no credit card required.

Data handling and privacy:

• Scope of data accessed: the app reads ticket metadata and image attachments (content URLs, filenames, comment IDs) from your Zendesk account via OAuth. It does not read or store ticket text, requester personal data, or non-image attachments.

• How images are processed: image bytes are fetched over HTTPS from Zendesk through your authorized OAuth token and sent to the Sentinel detection API for analysis. Images are processed in memory and discarded after the scan completes; only the resulting risk score, indicators, and a SHA-256 hash (used to deduplicate re-scans) are retained.

• Storage and retention: scan results are stored in Sentinel’s EU-hosted database, linked to your organization account, and are accessible only to authenticated members of your Sentinel organization. You can delete your organization and all associated scan data at any time from the Sentinel dashboard.

• Security: all traffic uses TLS 1.2+. Authentication is performed via OAuth (for Zendesk) and passwordless magic links (for Sentinel accounts). No passwords are stored. Access to stored scan data is restricted to the organization members you invite.

• Unauthorized access protection: access to the Sentinel API requires a valid, scoped token issued to the authenticated user; tokens can be revoked at any time by disconnecting the app or signing out. Sentinel follows least-privilege access controls and logs administrative access to customer data.

• Your rights: you can request export or deletion of your data at any time by contacting support@tarczasentinel.pl. Full details are available in the Privacy Policy.