Tenna Systems is applying vertical data moats to cybersecurity, representing a seed vertical AI play with enhancement generative AI integration.
As agentic architectures emerge as the dominant build pattern, Tenna Systems is positioned to benefit from enterprise demand for autonomous workflow solutions. The timing aligns with broader market readiness for AI systems that can execute multi-step tasks without human intervention.
Tenna Systems provides software solutions focused on electromagnetic spectrum intelligence for defense and critical mobility sectors.
The ability to fuse telemetry from heterogeneous, already-deployed receivers into a confidence-scored, real-time geolocation and interference picture using geometry-driven algorithms plus ML — effectively turning each operational platform into a sensor and offering EW resilience without hardware replacement.
Tenna aggregates proprietary, domain-specific RF telemetry and operational signals from commercial and military sources (including active operational deployments). This dataset scope and privileged access form a vertical, industry-specific data advantage used to train models and provide differentiated spectrum intelligence.
Unlocks AI applications in regulated industries where generic models fail. Creates acquisition targets for incumbents.
The stack appears to use a collection of specialized algorithms and models (classical signal-processing modules, statistical estimators, and purpose-built ML classifiers/regressors) routed to specific tasks (detection, classification, geolocation, confidence scoring). Different components (Arena, Tracer, Halo) imply task-specific modelization rather than a single monolithic model.
Cost-effective AI deployment for mid-market. Creates opportunity for specialized model providers.
Tenna's architecture implies streaming ingestion and operational feedback from distributed sensors (crowdsourced and partner platforms). This enables iterative retraining/updates driven by new telemetry and operational outcomes—a feedback loop that continuously improves models and detection/geolocation accuracy.
Winner-take-most dynamics in categories where well-executed. Defensibility against well-funded competitors.
There is a distinct validation/decisioning layer that assesses incoming signals and model outputs for trustworthiness (reject spoofed signals, provide confidence scores). Functionally this acts as a guardrail/verification model or rule layer that filters or corrects upstream inferences before acting on them.
Emerging pattern with potential to unlock new application categories.
Hybrid orchestration: heterogeneous telemetry ingestion -> cloud-scale fusion & ML inference -> task services (geolocation) -> downstream policy/actuation via middleware (Halo) on devices. Orchestration likely uses microservices and APIs rather than multi-model LLM chains.
Former Space Program Director, Operational Requirements Department at the Israeli Air Force
Previously: Israeli Air Force
Twin brother of Avner; previously led signals-intelligence and electronic-warfare programs within the Israeli defense ecosystem
Previously: Israeli defense / Israeli military
high
partnership led
Target: enterprise
custom
field sales
• Engagements with U.S. Army and U.S. Air Force
• Collaboration with Israeli Ministry of Defense
• Operational deployments in contested environments
Provide real-time spectrum intelligence by detecting, classifying, and geolocating RF interference and converting existing sensors into a unified spectrum picture for threat awareness and decision support
The tight product trio (middleware that upgrades legacy sensors + cloud fusion + precision geolocation algorithms) that explicitly avoids new hardware offers an unusual, software-first route to electronic warfare capabilities.
Embedding automated signal acceptance/rejection policies directly into legacy receivers (without hardware changes) is an impactful, operationally oriented safety approach uncommon outside specialized EW fields.
Tenna Systems operates in a competitive landscape that includes Raytheon Technologies / RTX (including Collins Aerospace), Northrop Grumman, BAE Systems.
Differentiation: Hardware- and platform-centric prime contractor with integrated EW suites; Tenna is software-first and hardware-agnostic, emphasizing rapid software integration onto existing receivers rather than new sensor platforms.
Differentiation: Focuses on end-to-end systems and proprietary sensors; Tenna competes by converting already-deployed receivers into a distributed sensing fabric and offering faster, lower-cost deployability via software.
Differentiation: Primarily sells hardened, hardware-integrated EW solutions and upgrades; Tenna sells software middleware and cloud/edge services that augment legacy receivers without rip-and-replace procurement.
Software-first, sensor-agnostic TDOA/network-geolocation focus: Tenna emphasizes geometry-driven geolocation (TDOA and multilateration) using already-installed receivers rather than traditional signal-feature/ESM analysis. That flips the usual SIGINT tradeoff — prioritize spatial fusion and timing geometry over expensive RF front-end fidelity.
Middleware 'Halo' that retrofits legacy GNSS/receivers in-field: Claiming to embed decision logic into existing receivers (reject/spoof decisions, share threat intel) without hardware swaps implies deep integration pathing (firmware hooks, OEM partnerships) and an on-device policy/validation layer — a non-trivial product that operates inside safety-critical stacks.
Cross-domain, multi-platform sensor fusion at internet scale: They ingest airborne, space, and terrestrial telemetry into a unified, real-time spectrum picture. Combining extremely heterogeneous sampling rates, observability, and latencies (satellite passes, airborne receivers, ground COTS) requires bespoke temporal alignment, interpolation, and uncertainty propagation.
Operational datasets from contested battlespace as training signal: Tenna claims real-world combat provenance and models trained on 'billions' of data points — if true, that provides labeled, adversarial examples (jamming/spoofing signatures and geolocation labels) that are extremely rare commercially.
Confidence-scored, tactical CEP outputs (50–200m): They emphasize producing probabilistic geolocation (with confidence scores) fast enough for operational decisions — implying fast Bayesian/likelihood fusion, real-time filtering, and geometry-aware uncertainty modeling rather than batch analytics.
Tenna Systems's execution will test whether vertical data moats can deliver sustainable competitive advantage in cybersecurity. A successful outcome would validate the vertical AI thesis and likely trigger increased investment in similar plays. Incumbents in cybersecurity should monitor closely for early signs of customer adoption.
“combining global sensor data with AI models trained on billions of data points to deliver a spectrum picture that sees farther and responds faster than any standalone system.”
“AI models trained on billions of data points”
“Geometry-driven, sensor-agnostic geolocation emphasis (prioritizes geometry/TDOA and distributed sensor geometry over signal-level modulation/ESM details).”
“Hardware-free, middleware-first deployment (Halo) that embeds resilience and trust logic into legacy receivers enabling both on-device and remote workflows without hardware swaps.”
“Crowdsourced sensor model: treat every connected device/receiver as a potential telemetry source to produce a persistent, ubiquitous spectrum picture.”
“Hybrid classical+ML stack: explicit combination of deterministic signal-processing techniques (TDOA, statistical estimation) with ML classifiers/regressors and operational confidence scoring.”
