Skip to content

Cryvanta AI SOC

Early access

An analyst desk that never clocks off

Cryvanta's AI analyst triages, investigates and rules on detections around the clock — and shows you every step it took, and every time it chose to hold. It's in early access with design partners today, EU-hosted, and runs on the model you choose.

Stage
Early access
Sensor
Windows today
Hosting
EU only
Model
Yours

The desk

Every verdict comes with its reasoning attached

A score tells you nothing you can act on. The analyst shows the events it correlated, the hypotheses it weighed, what it did — and where it stopped and waited for you.

Verdict recordWKS-0413 · 09:41 CET

Evidence

  • 09:41:02process.createwinword.exe → powershell
  • 09:41:02net.connect185.104.· · ·:443
  • 09:41:03file.write%TEMP%\up1.dat
  • 09:41:03registry.set…\Run\OneDriveSyncer
  • 09:41:04process.createrundll32.exe up1.dat
Detection

PowerShell download cradle

T1059.001

Analyst reasoning

  • Correlated 5 events on WKS-0413 within 2s
  • Parent chain matches T1059.001 download cradle
  • No change ticket, no maintenance window
  • Host holds finance shares — blast radius high
VerdictTrue positive · host isolated
Disk wipe requires human approval — held
Illustrative record · not live data

The night shift

Nobody reads forty thousand events

Alerts do not stop at 5 p.m., and the queue that greets a human at 09:00 is mostly nothing. The work worth a person's attention is the last line of this list — and finding it is the job.

One night · one estate

18:00 → 06:00 CET

  1. 41,900

    Endpoint events

    Everything the sensors saw across the estate between 18:00 and 06:00.

  2. 312

    Rule matches

    What the Sigma corpus flagged, at the edge and again in the cloud.

  3. 46

    Correlated findings

    Matches stitched into incidents per host, so one intrusion is one item.

  4. 5

    Escalated to a human

    Cases the analyst could not close on the evidence it had, with its reasoning attached.

  5. 1

    Held for approval

    An irreversible action the analyst wanted to take and would not take alone.

Illustrative night · not live data

Capabilities

The reading, the reasoning, the response

Alerts don't stop at 5 p.m., and nobody can read every one at 3 a.m. Cryvanta works the queue at any hour — and leaves the judgement calls a human wants to keep.

01

Triage every finding

The analyst reads each detection with the events around it and decides whether it's real, benign, or worth a human's time — with a rationale attached, not just a score.

02

Investigate the hard ones

When a finding is ambiguous, it gathers the surrounding activity, weighs competing explanations, and writes up its reasoning before it commits to a verdict.

03

Contain, when you allow it

It can quarantine a file or stop a process — but irreversible actions hold for a human unless you've explicitly told it otherwise.

04

Sharpen its own detections

Patterns it sees repeatedly become proposed detection-rule changes, compile-checked and put in front of an operator before they go live.

What's live

What's working, and what isn't yet

Early access means genuinely useful and genuinely unfinished at the same time. Here's the honest split so you can decide if now is the right time to start with us.

Live today

  • AI analyst: triage, investigation and verdicts
  • Windows endpoint sensor + cloud pipeline
  • Safety brakes, most shipping in shadow mode
  • Multi-tenant console with per-tenant isolation
  • GDPR erasure & audit-log retention controls

Ahead of us

  • macOS and Linux endpoint sensors
  • General availability beyond design partners
  • SOC 2 and independent certification
  • Self-service tenant onboarding

The safety brakes and the pipeline are covered in depth on the platform page.

Sovereignty

Your model, your region, your ceiling

Your model

Run the analyst on Anthropic, or an EU-hosted model that meets your residency rules.

Your region

Telemetry, findings and history stay in the EU. Nothing has to leave the continent.

Your budget

A hard spend cap sits in front of every model call, so autonomy never becomes an open tab.

Design-partner programme

Shape it while it's being built

We're taking on a small number of EU organisations as design partners. You get the platform early and a real say in where it goes.

Direct line to the builders

You work with the people writing the code. Feedback lands in the product in days, not quarters.

Your scenarios, prioritised

We tune the analyst and the detections against the threats that actually matter to your estate.

Honest reporting throughout

You'll always know which behaviours are enforced, which are shadowed, and what's still on the bench.

Join the early-access programme

If you're an EU organisation weighing up AI-native detection and response, we'd like to hear what you're up against. No pressure, no oversell.