AI-powered email security

Modern email security that learns as it works.

Cloud-based mail security gateway powered by machine learning. Stops spam, phishing and malware in under 5 milliseconds — with no hardware to deploy and no capital investment. Every message is scored by a custom-trained classifier and, where the verdict is uncertain, escalated to Claude AI for a second opinion. Bad mail is rejected, quarantined or tagged before it reaches your inbox, while legitimate mail flows through untouched.

app.mxguard.uk · live feed
Live
Total
2,847
Clean
2,521
Quar.
214
Reject
112
TimeFromSubjectScoreVerdict
10:32bounces@boden.comYour order has shipped0.04Clean
10:30jszfkprtq@xyz.clubRe: invoice 44210.87Quar.
10:30admin@phish-bank.clickUrgent: verify account0.99Reject
10:29📬 receipts@stripe.comPayment received0.01Clean
10:28winner@lottery-uk.winCongratulations winner!!!0.62Tag
10:27orders@amazon.co.ukOrder dispatched0.03Clean
10:26support@github.com[mxguard] new pull request0.01Clean
Slide 1 of 8 · Live feed
Message detail — 0.99 score
From: admin@phish-bank.click
Subject: Urgent: verify your account
To: accounts@acme.co.uk
ML model
0.95
Claude AI
phish
URL rep
URLhaus
Sender rep
0.78
SPF/DKIM
fail
Heuristics
+0.30
Final 0.99 — rejected at SMTP. URLhaus + Claude=phish + .click + gibberish.
Slide 2 of 8 · Message detail
MX Node Health
refresh 30s
mx1.mxguard.uk
online
194.26.222.170 · UK
scannerpostfixdkimarcclamav
Postgresprimary
Redismaster
Queuehold 291
Load0.19
mx2.mxguard.uk
online
178.105.120.61 · Helsinki
scannerpostfixdkimarcclamav
Postgresstandby 0.8s
Redisreplica
Queuehold 0
Load0.00
Active-active HA. PostgreSQL streaming & Redis replication over mutual TLS. Either fails, mail keeps flowing.
Slide 3 of 8 · Node Health
Customer dashboard — today
Total
487
Clean
412
Quar.
38
Reject
37
7-day trend
Top recipients
info@acme · 142
sales@acme · 98
Top blocked
*@evil.com · 87
*@scam.click · 54
Slide 4 of 8 · Dashboard
Quarantine — 12 messages held
TimeFromSubjectScoreAction
10:32noreply@invoice.zipInvoice please pay0.91
10:18kxjphw@quotes.workRe: project quote0.84
09:47news@listings.coNewsletter digest0.62
09:12winner@lottery.winCongratulations!0.72
08:55offer@deals.shop70% discount0.81
08:31support@gen-bank.coAccount locked0.88
Retention 30 days. Releases retrain the ML model.
Slide 5 of 8 · Quarantine
Allow/block rules — 4 scope tiers
ActionPatternScopeHits
block*@evil.comglobal2,341
block*@spammer.shopcustomer187
allow*@hetzner.comglobal94
allowmarketing@trusted.comdomain42
blocknoreply@badco.comrecipient11
recipient
one mailbox
domain
one domain
customer
all yours
global
admin only
Slide 6 of 8 · Rules
Add a domain — self-service
Domain
acme-ltd.co.uk
Destination mailserver
mail.acme-ltd.co.uk:25
Step 1 · verify ownership
_mxguard-verify.acme-ltd.co.uk TXT "mxg-9f2a…"
Step 2 · point MX records
acme-ltd.co.uk. MX 10 mx1.mxguard.uk.
acme-ltd.co.uk. MX 20 mx2.mxguard.uk.
Step 3 · traffic flows automatically
live
Slide 7 of 8 · Add domain
Your domains — 8 active
DomainBackendStatusToday
acme-ltd.co.ukmail.acme-ltd:25active142
acme.ukmail.acme-ltd:25active88
acme.commail.acme-ltd:25active61
acmegroup.iomail.acme-ltd:25active37
acme.orgmail.acme-ltd:25active22
acmecorp.netmail.acme-ltd:25pending
Showing 1–6 of 8
Slide 8 of 8 · Domains
Auto-rotating preview · sample data app.mxguard.uk

AI detection

  • +Custom ML classifier scoring at AUC 0.988
  • +Claude AI second-pass on uncertain mail
  • +LightGBM gradient-boosted classification
  • +Scores subject, body, headers and links together
  • +Attachment and URL risk analysis

Speed & resilience

  • +Sub-5ms scoring per message
  • +Two redundant nodes — UK + EU active-active failover
  • +No hardware or appliances to maintain
  • +No single point of failure
  • +Handles high-volume mail without queueing

UK & compliance

  • +UK-hosted, GDPR-aligned, no US data residency
  • +Over three decades of UK internet infrastructure experience
  • +Data processed and stored within the UK & EU
  • +Nominet-accredited UK registrar heritage
  • +Full audit trail of every decision

Transparent & adaptive

  • +Transparent verdicts — no black box
  • +Self-learning from your ham/spam labels
  • +See exactly why each message scored as it did
  • +One-click mark-as-ham, allow or block
  • +30-day quarantine with self-service release

Email security trusted by businesses across the UK and globally.

From a 33-year-standing company.

99.5%
Detection accuracy
<5ms
Per-message scoring
3.1B
Messages analysed yearly
24/7
Cloud availability

Why MxGuard

Built for the threats your business actually faces.

Traditional anti-spam tools were designed decades ago. MxGuard replaces them with a modern ML-first architecture — cloud-delivered, transparent, and self-improving.

Stop modern threats

Targeted phishing, BEC, malware attachments, credential harvesters — caught by an ML classifier (AUC 0.988) plus Claude AI second-pass on borderline cases.

Real-time scoring

Every message scored in under 5 ms. Watch mail flow live in the dashboard with full visibility into every verdict.

Transparent verdicts

Click any message to see exactly why it scored as it did. No black box — defensible decisions you can explain.

High availability

Two redundant nodes in the UK and EU running active-active with replicated state. If either node goes offline, mail keeps flowing automatically — no DNS change, no intervention.

Self-learning

Every ham/spam label your team adds improves scoring. The system adapts to your specific mail patterns over time.

UK-hosted, GDPR-aligned

All mail processing and storage in the UK. Built and operated by Transcom Internet Services — over three decades in UK internet infrastructure.

Try MxGuard free for 30 days.

Full feature access, no credit card required. Choose any plan level for the trial — we will email your access key in minutes.