Skip to content
banner-anomaly-detection

Behavioural detection
that stays on-site

Anomaly Detection is an add-on for Guardsix SIEM that finds what no rule was written for, entirely on your own infrastructure.

Catch the account behaving out of character, and see exactly why it was flagged, with reasoning your analysts and your auditor can follow.

The guardsix SIEM

Designed for the reality European teams live every day

join_green4_48px Stop threats like credential abuse without extra headcount  
join_green4_48px Faster investigations without complex querying  
join_green4_48px Compliance readiness built into daily operations  

Behavioural analytics should not require sending your logs off-site

Many cloud-first SIEMs deliver user and entity behaviour analytics as a service in the vendor's infrastructure. The logs that describe how your people and systems behave have to travel there to be analysed. For lean teams at regulated organisations, that raises three obstacles:

Logs that leave your jurisdiction

Behavioural baselines are built from who signed in, when, from where, and what they touched. Sending that stream to a vendor's cloud takes it outside your jurisdiction.

A bill that follows your data

Cloud analytics are often priced by the volume of data you send. A new log source, an incident or a longer retention period raises the bill at exactly the moment the team leans on the tool hardest.

A score without a reason

When a tool returns a risk score without its baseline or its reasoning, your team stays accountable for a decision it cannot explain to an auditor.

people-society-23-medium

Explainable anomaly detection, on your own infrastructure

Anomaly detection learns what normal looks like for each account and system, then flags behaviour that departs from it: a login at an hour that person never works, an action repeated far more often than usual, or a sudden rise in data transfer.

How it works

Run four detectors on log sources across your estate:

arrow_forward-24px
Entity attribute rarity
arrow_forward-24px
Entity frequency drift
arrow_forward-24px
Time window deviation
arrow_forward-24px
Statistical measure

Each score uses statistical profiling a person can read and adjust, so there is no opaque model standing between analysts and answers.

Illustrative example Statistical measure
svc-backup · finance file share 82
Observed
Read 412 files on the finance share in one hour
Baseline
Under 20 files a day for this account and its peers
Why it scored
Rare for this entity and its peer group, on a high-criticality share
MITRE ATT&CK
Collection

See why every anomaly was flagged

Every anomaly carries four things: what was observed, what the baseline was, why it scored, and the score itself. The reasoning reads the same way across every detector, so an analyst opens a lead with a place to start rather than a number to decode.

When the auditor asks how you monitor risky activity, the evidence is already there, searchable and traceable, in its own repository. The analyst keeps the decision. The platform shows its work.

Detection you can explain, running where your logs already are

Anomaly Detection runs on Guardsix SIEM, with the sovereign-by-design architecture and node-based pricing teams already rely on. It is sold and delivered through Guardsix partners, so you work with people who know your environment.

 

  • Coverage without extra headcount

    Behavioural baselines surface the unusual, including patterns nobody wrote a rule for.

  • Reasoning your auditor can follow

    Every anomaly records what was observed, the baseline, why it scored, and the score.

  • Your data stays under your law

    Detection runs entirely inside your deployment, air-gapped estates included.

  • Spend you can plan

    Priced as a licence, with no ingestion or per-query metering behind it.

Quote from RENK on the Anomaly Detection add-on

Gradient light mesh g9 1

Behavioural detection you can explain to an auditor