What changes for the organisation
Localised, multilingual scams that generic filters miss get caught and explained.
Reads inbound mail and malware behaviour for the regional attacks generic filters miss.
What changes for the organisation
Localised, multilingual scams that generic filters miss get caught and explained.
How it works
Inbound mail is analysed for impersonation, social engineering, suspicious links and attachments, QR phishing and conversation hijacking, alongside mail-authentication, reputation and sandbox signals. Sandbox behaviour, imports, strings and decompiler output are summarised into candidate indicators and draft detection rules.
The mechanism, not the marketing. If a step below would not survive an engineer's scrutiny, it should not be on this page.
Operating path
Authentication passed. No malicious link. No attachment. Scored clean by every conventional control.
A message in Bahasa Indonesia referencing a real, open purchase order by number.
Legitimately — the sender's domain was registered nine days ago and configured correctly. Authentication proves control of a domain, not honesty.
Matches a known supplier contact exactly. The underlying address does not.
The message claims continuity with a conversation that does not exist anywhere in the mail history.
Changed bank details, and an urgency construction idiomatic to the language — the kind of pressure phrasing that does not survive translation and therefore does not appear in an English-trained filter's training data.
Impersonation flagged with all five supporting signals cited, provenance separated between authentication result, reputation, mail history and model inference.
The finance team calls the supplier on a number from their own records — not one in the email.
What it came to
The payment did not go out. Every conventional control had scored this clean, and each of them was working correctly: the attack was carried in language and in conversational context, not in a link or a file.
The authority ladder. Green rungs are authoritative — a rule, a scanner, a sandbox, or a person. Blue is where the model contributes. Amber is where a human decides.
Language and regional judgement — the Bahasa-Indonesian invoice scam or the Thai-language impersonation that an English-trained filter reads as ordinary business mail.
It cannot determine that an artifact is malicious on its own; the sandbox and the analyst do that.
We will not show you a scripted demonstration of something that does not exist. What we publish instead is the workflow specification, held to the same standard as shipped work: actor, trigger, happy path, exceptions, terminal state, and the audit evidence it must leave. When it is built, this section becomes a demonstration.
Roadmap discovery. Promotion requires a named mail platform and sandbox, plus per-language detection-quality measurement.