Beyond spam: classification against fraud.
Spam is an annoyance. Business email compromise is an existential threat — the FBI logged $2.77 billion in BEC losses in 2024 alone. BORGPost was designed for the second problem.
The dangerous email looks clean.
The message that empties a bank account carries no virus and trips no spam filter. It’s a polite note from a “supplier” whose domain is one character off. It’s a reply inserted into a genuine thread. It’s your “CEO” asking accounts to move fast on a wire transfer. Filters built to score junk pass these through, because there’s nothing junky about them — the fraud is in the context, not the content.
BORGPost reads context. Every message is examined the way a security analyst would examine it: who is this really from, does the domain check out, do the links go where they claim, does the request pattern match known fraud playbooks?
What classification actually means
Named threats, not scores
Thirteen categories including business email compromise, spear phishing, domain and brand impersonation, conversation hijacking, account takeover and extortion. Your team learns to recognise the pattern because the system names it.
Reasoning you can read
Each verdict — Safe, Suspicious or Malicious — comes with a confidence score, the red flags found and a plain-English explanation, shown right in the message. Suspicious verdicts open the panel automatically.
Action, not just alerts
Every message gets a recommendation — deliver, flag or quarantine — and your server-side rules can act on the classification automatically. Admins see threat activity across the whole organisation.
Defence in depth, on by default
Detection is one layer. Underneath it, every email is sanitised through a strict allowlist before it renders — no scripts, no active content. Remote images are blocked until you choose to load them, so tracking pixels report nothing. Inline images and attachments are served through signed, expiring tokens. Links are screened for look-alike and homograph tricks before you click. And the classifier itself is hardened against prompt injection: instructions hidden inside a message are never followed — they’re flagged as an attack.
Your domain, authenticated properly
Fewer than one in six domains publishes any DMARC policy at all — which is why criminals can spoof most small businesses freely. When BORGPost hosts your domain, SPF, DKIM and DMARC are configured as part of onboarding, on infrastructure with a clean, dedicated sending reputation. Your invoices reach inboxes; forgeries of your name don’t.
Ownership without the burden
Your mail lives on infrastructure run for you, not mined by an advertising business. Sign-in is passkey-first with TOTP and backup codes; stored credentials are encrypted at rest; sessions and APIs are hardened against the usual web attacks. The intelligence engine is provider-pluggable — including locally hosted models — so analysis can run without your mail leaving the server. Standard IMAP and SMTP mean you can leave any time. We’d rather keep you with the product than with a lock.