Customers · Marketplace
Marketplaces sit on the largest behavioral dataset in fraud. The blocker is labels. Roe reads chats, listings, and transactions. Returns typology-tagged labels for every closed case.
What you get
Every for marketplaces deployment ships with the same evidence-first reasoning, audit trail, and policy fit as the rest of Roe.
Off-platform redirect, fake refund, romance, employment, ticket, rental. Every confirmed scam ships with the variant tag your classifier needs.
Reads chat content, listing language, and image-based phishing. Distinguishes credential phishing from scam-payment phishing from impersonation.
Links accounts, devices, IPs, and behavioral clusters. Returns ring-level labels so your model learns the whole network shape rather than a single account.
Backfill historical cases or stream new labels as they close. Lands in your warehouse next to the features that drove them, ready for the next training run.
Scam and phishing typologies
Labels with cited evidence
Lands beside the features that drove them