They're listening.
We tell you when, from where, and what to do about it.

In the Pegasus era, any protected person — journalist, executive, diplomat, case officer — walks through hostile cellular surveillance they cannot see. IMSI catchers, rogue base stations, SS7 exploits, silent SMS pings, encryption-downgrade attacks. Bastion is a family of passive cellular monitors that detect, classify, and locate hostile intercept gear operating against a protected asset. It never intercepts anything. It only tells you when something is looking at you.

Cellular is the invisible tradecraft layer nobody defends.

Every protected person carries a cellular device. That device is the single largest untrusted surface in their operational posture. The post-Pegasus disclosures (2021–2024) made one thing clear: the intercept ecosystem is global, industrialized, and targets civilians as routinely as it targets military operations.

A passive listener that never intercepts.

Bastion is counter-SIGINT at the cellular layer. It watches the cellular RF environment around a protected asset and flags hostile intercept gear in real time. It is the ground/personal counterpart to Nexus Atlas Blackbird (airborne emitter detection) and Phantom (RF deception). The three together constitute our counter-ISR family.

Three tiers. One protected person, one vehicle, one site.

A single product architecture scaled across three physical form factors. Shared detection stack, shared library, shared backend. Different antenna arrays, different channel counts, different customer segments.

Ten classes of indicator. Correlated, weighted, confidence-scored.

No single indicator is conclusive. A rogue base station typically trips five to eight indicators simultaneously. Bastion's correlation engine fuses weighted rule hits with baseline-anomaly scores and RF-fingerprint confidence to produce one of four states: Green Amber Red Critical

Know who's watching.

The signature library is built backwards from known adversary hardware. Each catcher class has a distinctive RF fingerprint footprint, a characteristic signaling behavior, and a typical deployment pattern. Cataloging adversary variants is the research heart of the product.

Antenna to alert ladder in under two seconds.

Passive RF capture, baseband DSP, per-standard protocol decoding, three-pipeline detection engine, correlation, alert ladder. All on-device for privacy; only alert metadata and flagged IQ snippets leave for backend library updates.

  ┌──────────────────────── BASTION-S · SIGNAL PATH ───────────────────────┐
  │                                                                        │
  │    ANTENNAS                                                            │
  │    ┌─────────┐  ┌──────────┐                                           │
  │    │ WIDEBAND│  │ 5G PATCH │  ← polarization-diverse                   │
  │    │ 0.6–3GHz│  │ 3.3–6GHz │     internal to unibody                   │
  │    └────┬────┘  └─────┬────┘                                           │
  │         │              │                                               │
  │    ┌────▼──────────────▼────┐                                          │
  │    │ LNA + FILTER BANK      │ ← SAW filters, 5 cellular bands          │
  │    │ (band-switched)        │                                          │
  │    └────────┬───────────────┘                                          │
  │             │                                                          │
  │    ┌────────▼─────────┐                                                │
  │    │ SDR · AD9361     │ ← 70 MHz – 6 GHz, 56 MHz IBW, 12-bit           │
  │    │ 2×2 MIMO, 12-bit │                                                │
  │    └────────┬─────────┘                                                │
  │             │  IQ samples (DMA)                                        │
  │    ┌────────▼─────────┐                                                │
  │    │ BASEBAND DSP     │ ← sync, channel est., demod                    │
  │    └───┬────────┬─────┘                                                │
  │        │        │                                                      │
  │   ┌────▼───┐ ┌──▼────────┐ ┌──────────┐                                │
  │   │PROTOCOL│ │RF FINGER- │ │ RAW-IQ   │  ← 24h rolling triage buffer   │
  │   │DECODER │ │PRINT      │ │ BUFFER   │                                │
  │   │GSM·UMTS│ │EXTRACTOR  │ │          │                                │
  │   │LTE·5G  │ │(32-dim)   │ │          │                                │
  │   └────┬───┘ └──┬────────┘ └──────────┘                                │
  │        │        │                                                      │
  │   ┌────▼────────▼────────┐                                             │
  │   │ DETECTION ENGINE     │                                             │
  │   │ rules + ML + baseline│                                             │
  │   │ correlation          │                                             │
  │   └───────────┬──────────┘                                             │
  │               │                                                        │
  │   ┌───────────▼──────────┐                                             │
  │   │ ALERT LADDER         │ Green · Amber · Red · Critical              │
  │   └───┬──────────────┬───┘                                             │
  │       │              │                                                 │
  │       ▼              ▼                                                 │
  │    HAPTIC           BLE → paired app / earpiece                        │
  │                                                                        │
  └────────────────────────────────────────────────────────────────────────┘

                         ▼ (tethered upload)
           ┌──────────────────────────────────────┐
           │  BACKEND — threat-intel + library    │
           │  ├ analyst-reviewed novel anomalies  │
           │  ├ operator fingerprint DB           │
           │  ├ IMSI-catcher variant library      │
           │  └ signed OTA library push to fleet  │
           └──────────────────────────────────────┘

The library would be the product.

Hardware commoditizes in 24 months. The library does not. Two families of signatures form the moat: operator RF fingerprints (what real cells look like at RF, per operator, per region, per vendor) and IMSI-catcher variant library (what known hostile hardware looks like). The former is huge and slow-building; the latter is smaller and evolves weekly as adversary kit is captured, analyzed, or red-teamed.

Every real BTS has an imperfection fingerprint.

A real cellular base station is built from production hardware — Huawei BTS3900, Ericsson RBS 6201, Nokia Flexi, ZTE Wireless. Each unit has characteristic RF imperfections that don't show up in the protocol but are measurable in the IQ samples. A rogue base station built from USRP / BladeRF / purpose-built catcher hardware leaves different imperfections. Match vectors against the operator library; mismatches are the single most reliable detection signal.

Illustrative. No system built yet. This is the mission we would solve.

A journalist meets a sensitive source in Istanbul. The local intelligence service has an interest. The meeting venue is a café on a busy street — legitimate cells everywhere, any of which could be cloned.

Nine segments. All underserved today.

Bastion's buyer base is broad because the threat is broad. What these segments share: high-value individuals or sites, credible cellular-surveillance exposure, and no incumbent vendor serving their price band.

The adversary adapts. So does the library.

Catchers evolve. Adversaries retune hardware, rotate software, update protocols. Library refresh cadence is the product's operational promise. A library six months stale is a library that's being defeated in the field. This is why the subscription model is the actual business — not the hardware.

A captured device yields nothing.

Bastion devices are carried by high-threat individuals. Device capture is a realistic scenario. Every design decision assumes the device will eventually be in adversary hands.

A barbell market. Nothing in the middle, where the buyers are.

Cellular counter-surveillance has two existing ends: enterprise ($500K+ Delma, Airbus, R&S) and consumer (free: EFF Crocodile Hunter, SnoopSnitch, Amnesty MVT). Between them — pocket-size professional-grade devices at $2–10K — there is nothing. That's where Bastion lives.

Nobody does what Bastion would do.

Don't attack the top. Disrupt from below.

Competing head-on with Delma, Airbus, or R&S in the enterprise market would fail. Their incumbency is 10+ years deep; their certifications are a 3–5 year moat; their export channels are pre-cleared; their procurement relationships are entrenched. A Bulgarian concept-stage startup cannot displace that on day one. The strategy is sequential.

Three products. One counter-ISR doctrine.

Bastion completes the Nexus Atlas counter-ISR family. Each product addresses a different layer of the adversary's intelligence cycle; together they form a closed loop with a shared signature-library moat.

  ┌──────────────── NEXUS ATLAS COUNTER-ISR FAMILY · CLOSED LOOP ────────┐
  │                                                                      │
  │       BLACKBIRD                 PHANTOM                BASTION       │
  │    (airborne detect)         (RF deception)       (cellular detect)  │
  │           │                       │                        │           │
  │           ▼                       ▼                        ▼           │
  │     adversary drones          adversary SIGINT          hostile cellular │
  │     + radar emitters          cycle corrupted          intercept gear   │
  │           │                       │                        │           │
  │           └──────── shared library moat ───────────────────┘           │
  │                  operator + adversary                                │
  │                  RF fingerprint database                             │
  │                                                                      │
  │       ┌── Bastion detects hostile catcher ──────────┐                 │
  │       │  → target emitter added to Blackbird list   │                 │
  │       │  → Phantom can fake cellular if desired     │                 │
  │       └──────────────────────────────────────────────┘                 │
  │                                                                      │
  └──────────────────────────────────────────────────────────────────────┘

Counter-surveillance, not interception.

The single most important strategic point about Bastion: it sells as counter-surveillance equipment, not interception equipment. This is a clean ECCN 5.A.1 counter-surveillance category — not on the EU Dual-Use 2021/821 Annex I 5.A.1.f interception list.

Concept phase. Pre-prototype. Idea presented for partner interest.

Bastion is currently a concept being presented for early partner interest. Nothing is built or deployed. The roadmap below is what we would execute, given funding and partnership.