Sharper spam filtering. Tuned to you.
MXFilter catches 99%+ of spam, malware, and phishing, with per-domain Bayesian learning that sharpens based on your messages, not someone else's. Built in-house. Over 1 million messages scored daily across our network.
- 99%+ catch rate
- <0.1% false positives
- Tuned to your inbox
- DMARC + MTA-STS included
Always evolving.
The filter protecting you today isn't the one you had last month. The engineers who built it keep sharpening it, against new threats and your own false positives.
Always tuning
Report a false positive and the engineers who wrote the filter see it. Tunings go live same day, sometimes same hour. Not on a vendor's SLA.
Tuned for you
Per-domain Bayesian learning kicks in once a user trains enough ham and spam. Your inbox learns your conversations, not the global average.
AI-augmented
Beyond rules and Bayes, a local LLM reads message intent to catch novel phishing and social engineering. Runs in-house, so your mail never leaves our network.
From simple matches to deep regex.
Build simple rules to catch domains and senders, or complex regex that matches anywhere in the email body and headers.
- Customize the blocking action (hard reject or quarantine)
- Match both sender & receiver
- Full raw message matching
- Intuitive regex tester (you'll need it)
Customized to your inbox.
Most filters share one Bayesian brain across every customer. We don't. Once a user trains enough ham and spam, MXFilter spins up a per-domain Bayesian model that sharpens against your real conversations.
- Activates automatically once trained on enough mail
- The longer you use it, the sharper it gets
- Falls back to our global model until then, so there's never a cold start
Tune scores where they matter.
Every verdict carries a score behind the scenes, and you can move those numbers. Loosen thresholds so newsletters stop getting eaten. Tighten them on a domain getting hammered by campaigns.
- Adjust account-wide or down to a single domain
- Tweak individual rule weights without touching defaults
- Live preview of any change before it hits the queue
Quarantine, reports, and digests.
Held mail lands in a searchable quarantine where users can preview and release it without filing a ticket. Scheduled digests show each person what got held, with one-click release, so most never log in at all.
- Full quarantine with search, preview, and release
- Report builder for traffic, scores, and threats
- Daily/weekly digest emails with one-click release
Spoof-proof your domain
DMARC reports show you exactly who's sending mail as you. MTA-STS makes sure inbound mail arrives encrypted, enforced rather than best-effort. Both come standard.
No single point of failure.
Four MXFilter gateways on three continents, each one able to take all of your inbound mail on its own. Every site runs the local LLM on its own RTX 5090s, so filtering happens at the edge and inbound stays online, even if one country goes to war.
Las Vegas
Nevada, USA
mta01.nathanael.llcLocal LLMAmsterdam
Noord-Holland, Netherlands
mta02.nathanael.llcLocal LLMNew York
New York, USA
mta03.nathanael.usLocal LLMSingapore
Singapore
mta04.nathanael.usLocal LLM
- No single point of failure
- Inbound always online
- Filtering at the edge
- Your mail never leaves our network
- Privacy? That's us.
Everything you'd expect, then some.
Allow & block lists
Per-user and per-domain sender allow/block lists with regex support.
Webhooks
Real-time message events for downstream tooling, SIEM, or custom workflows.
Dark & light mode
Full theme support across the panel.
Full archiving
Delivered, rejected, or quarantined, every email is archived and searchable.
TLS by default
STARTTLS and DANE on every connection, with per-message SPF and DKIM results.
Audit log
Every score override, allow-list change, and quarantine release is logged and searchable.
Spam, stopped.
Real engineers, real filtering, included free. MXFilter ships with every hosting product we sell. No add-on fees, no per-domain markup, nothing extra to buy.