Sovereign AI security that proves itself.

An AI analyst for your security operations center (SOC), continuous attack validation, and the platform under both — working with the endpoint, network and extended detection tools (EDR, NDR, XDR) you already run.

Every Phoenix product runs on your own hardware and reasons with a local, open-weights model you host. We don't ask you to trust the model. We show you the receipts.

customer logs sent to a hosted model 0
The problem every SOC already has

More signal than any human can read.

That isn't a defect — it's what a comprehensive ruleset does. The industry's fix is an AI analyst that reads the stream for you. Most hosted AI analysts work by sending your logs to someone else's cloud model.

01

Volume defeats attention

When a channel emits hundreds of alerts a day, it stops being read. Not through negligence — through arithmetic.

02

Rules only find the known

A detection rule is a hypothesis written in the past. It can't fire on the attack nobody wrote a rule for — and it can't tell you it didn't.

03

Quiet failures are the deadly ones

A SIEM that stopped receiving logs looks identical to a quiet night. So does triage that has been silently failing for days.

For regulated, sovereign and air-gapped operators, shipping logs to a cloud model isn't a preference — it's often prohibited outright, or gated behind approvals that never come. Phoenix is the other answer.

One incident, three lenses

Exposure → detection → response, without a byte leaving your walls.

Crucible proves the gap (attack) Nest detects it (respond) Core the platform both run on coverage_gap → detection improves
Why “proves itself” is not a slogan

Receipts, not claims.

Four principles, taken straight from how the software is actually built.

1

Authority lives in code, not the model

The model can name an action from a fixed list; it can never invent one. It can request a firewall block, but a check in code it can't see or change decides whether to run it. Assume it's fallible, misreading, even jailbroken: it can still only act inside that fixed, code-owned boundary. That holds across model releases.

2

Nothing is armed before it is measured

Every automated gate goes through three stages — watch, rehearse, then act (observe → shadow → enforce). It records the decision it would have made against live traffic before it is ever allowed to act on one.

3

Silence is the enemy, not error

The system is built to make its own failure noisy. A process that stops working is caught by a separate watchdog, because a stuck process can't raise its own alarm.

4

Provable non-egress

No customer log, alert or internal indicator leaves your network. Public indicators (a file hash, an external IP) are checked against reputation feeds only through one auditable chokepoint, and only for the feeds you switch on. It's enforced in code, and CI tests fail if the lookup paths are given a way around that chokepoint — an architecture you can audit and test behind your own deny-all firewall.

Who Phoenix is for

Built for the places a cloud AI can’t go.

Regulated finance & healthcare

Where log egress to a third-party model is a compliance non-starter.

Government, defense & OT

Air-gapped and critical-infrastructure environments the cloud cannot set foot in.

Any SOC drowning in alerts

That wants an AI first pass it can actually check, line by line.

About Phoenix

Phoenix builds AI security that runs on your own hardware.

Security teams get more alerts than anyone can read. The usual fix is an AI analyst in the cloud, which means sending your logs to someone else’s model. For banks, hospitals, government and critical infrastructure, that is often not allowed at all. Phoenix is built for them, and for any team that would rather keep its data at home.

What we make

  • Nest: an AI analyst for your security team. It triages every alert, reads the logs no rule matched, and hunts through days of data overnight.
  • Crucible: repeatable, authorized attack testing. It proves which weaknesses are real, and puts the ones attackers are already using first.
  • Core: the platform under both, connected to the Wazuh and OpenSearch tools you may already run.
  • Phoenix Apps: desktop and mobile apps that bring each person the alerts and fixes for their own devices.

Crucible attacks, Nest detects, and every gap it finds becomes a new detection rule.

What makes Phoenix different

  • Your data stays with you. The AI runs on an open-weights model you host. A cloud model is only ever an explicit opt-in. Outside lookups go through one auditable path, and our build fails if a change tries to go around it.
  • The AI never holds the keys. The model can only choose from a fixed list of actions. Code it cannot see or change decides what actually runs. New automation first watches, then rehearses, and only then may act.
  • Failures are loud. A separate watchdog catches a stuck process, because a broken monitoring system looks exactly like a quiet night.
  • We say where our limits are. A small local model is weaker than the big cloud ones. We say so openly, and we limit what it can do instead of asking you to trust it.

Receipts, not claims →

The bet behind Phoenix
The rule that customer data never leaves the building looks like a handicap. It forces a safety argument you can read in an afternoon — a worse analyst than a frontier model, and a better thing to put in front of a firewall. Read the full argument →

Ready to see the receipts?

[email protected]  ·  on-prem · local model · zero egress