Skip to content

Desktop application · Pre-release project

We see what
you don't.

Quiet, local-first defense for Windows. Every alert explained. Every action yours to confirm.

Illustrative dashboard preview · demo data

documented security detectors
14
where the optional AI model runs
127.0.0.1
fingerprints for critical files
SHA-256
confirm every destructive action
You

Capabilities

Visibility, with the reasons attached.

All features

Illustrations use demo data

How it works

From a signal to your decision.

Six steps. The last one is always a person — never an automatic destructive action.

  1. 01

    Observe

    Local system signals

  2. 02

    Detect

    Detectors & correlation

  3. 03

    Explain

    Evidence and reasons

  4. 04

    Investigate

    Controlled, read-only tools

  5. 05

    Recommend

    A proposal, with its limits

  6. 06

    Confirm

    A person approves

AEGIS · AI Security Analyst & Defense Agent

An analyst, not just a chatbot.

It reads this machine's telemetry through read-only tools, explains every finding, and proposes — never silently performs — a response.

  1. Rules decide

    Risk, confidence, verdicts and proposed actions come from readable, deterministic rules.

  2. Local model narrates

    Optional. It may narrate a decided report or pick one read-only tool. It never executes actions.

  3. You confirm

    A proposal is not an action. Nothing destructive runs until you approve it.

The AEGIS loop: Detect → Understand → Investigate → Recommend → Respond → Learn

Illustrative AEGIS panel · demo data · mockup control

Diagram: telemetry, AEGIS, and the optional local model at 127.0.0.1 sit inside your Windows PC. AEGIS does not use a cloud AI API. Optional notification channels can send information out only when you enable them.

Privacy

Local-first, by design.

AEGIS sends no telemetry to a cloud AI and makes no remote reputation lookups. Optional integrations only run when you turn them on.

Clear boundaries

What we won't claim.

A tool that hides its limits teaches you to trust it too much.

  1. 01Detection is not proof of malicious intent.
  2. 02Regular network patterns can be legitimate.
  3. 03Sampling can miss some fast activity.
  4. 04Access restrictions can leave results incomplete.
  5. 05Detecting every threat is not guaranteed.
  6. 06Pre-release until readiness is verified.

Know more about what your PC is doing.