About

Variene AI builds deployable edge computer vision for autonomous platforms operating in real, constrained environments. General purpose vision models are trained on broad datasets and best case conditions, so they fail quietly the moment they hit clutter, sensor noise, and partial visibility. That is the gap we close. Our mission is to give autonomous systems perception they can trust. We engineer purpose built architectures for the edge from day one instead of adapting cloud models, and our work covers four areas: detection and tracking of armored ground vehicles for pattern of life analysis; maritime awareness and vessel classification, including frigate versus non frigate; counter UAS and airspace awareness, including fast moving drones; and all weather synthetic aperture radar perception across maritime and terrestrial environments. We work in defense and dual use autonomy, a market growing fast as governments invest in sovereign capability and cheaper, software defined platforms. The problem is the same everywhere. Modern sensors produce far more data than any operator can process, and conventional systems break down exactly when conditions turn hard. Variene turns raw, multi sensor feeds into intelligence operators can act on immediately, at the edge, cutting their workload and closing the gap between detection and decision. Founded in Greece, Variene works with established defense primes and technology leaders to field perception built for deployment, not the lab.

Solution & Product

Variene's product is a family of perception models, each built for a specific sensor, target, and platform, not one model stretched across everything. The models fuse up to six sensor types in real time, electro optical, infrared, thermal, SAR, RF, and LIDAR, and run on the platform's own compute within tight size, weight, and power budgets. They output detections, classifications, and target tracks through standard interfaces, so they drop into an existing drone or unmanned system and feed its autonomy stack or the operator directly. Two things define the offer. First, it is edge native: inference happens on the device, so perception keeps working with no connectivity, no cloud, and no meaningful latency, exactly what a moving platform in a contested area needs. Second, it is sovereign: customers can train and run their own models on their own data, keep full ownership of that IP, and nothing leaves their infrastructure. The problem we solve is deployment. Most providers build for the cloud, where compute is unlimited and links are reliable, then compress that down to the edge after the fact. That path breaks on real hardware, and it forces sensitive data through someone else's systems. Our difference is that we start at the edge and at sovereignty from day one. We design each model for its target, sensor, and compute, then validate it in live operations, not lab benchmarks. The result is perception that ships, runs, and stays under the customer's control.

Innovation & Scalability

Variene's central innovation is a single edge native perception engine that spans six sensor types across four mission domains. Because those sensors and domains share one architecture, new capabilities reuse a common core and build on what already works. Most of the industry takes the harder path, standing up a bespoke, cloud trained system for every platform, which is slow and fragile in the field. Two design choices reinforce the approach. The technology is deployment first, built and tested in real operational conditions, and it is sovereign by design, so customers keep their own data and model IP. Variene's own edge compounds through a growing proprietary dataset, developed in part through sensor data partnerships, including for synthetic aperture radar. Scalability follows from software economics. A model built once for a sensor and target class deploys many times at low marginal cost, so revenue can grow well ahead of headcount. The company is deliberately asset light. It does not build airframes or sensors, and it is designed to integrate its perception layer into established platforms operated by partners and primes, so reach expands through partners while the internal cost base stays lean. The same shared engine means research effort accrues across the whole product line, so each improvement lifts several products at once. Growth comes from software, data, and partnerships, which scale far faster than people, hardware, or infrastructure

Traction & achievements

Proof of progress: customers, revenue, partnerships, awards, pilots, growth metrics. Founded in 2024, Variene reached its first defense programs and partnerships within roughly a year. Its flagship program is the ELKAK National Cargo UAV effort, a Greek initiative valued at around 25 million euros. Variene was selected for Phase 2 and leads the autonomy and perception work in the AERON consortium, alongside industrial partners including Bellavia, Autotech, Icomat, and Hoper. Variene's partnerships across defense and dual use include a publicly announced collaboration with Shield AI on autonomy integration, a synthetic aperture radar data partnership with ICEYE for maritime detection, and a partnership with Bertin. Its perception models have been tested in real operational conditions, including early field trials in Ukraine. The active pipeline includes the Hellenic Navy frigate program, a fit for Variene's maritime vessel classification work, an engagement with a North American counter UAS program as the threat assessment layer, and discussions with several international primes. Variene has presented at major industry events, including DEFEA in Athens, the Paris Air Show, and the Greek Defence Industry Compass in Brussels. The team has grown to around twenty people, and the company is in active discussions with institutional investors.

Team & Why Us

Variene is led by two principals. Stelios Koroneos, founder and CEO, set the company's sovereign autonomy thesis and leads its technology direction. He is an embedded systems engineer with depth in hardware, satellite communications, and wireless networks, alongside senior product roles in maritime data and vessel intelligence. Alex Moukas, President, is a serial entrepreneur and multi-venture operator who founded and led a technology company from inception through a NASDAQ listing and a follow on public offering, and has since built and scaled multiple ventures. They are backed by a team of around twenty across engineering, research, and business development, covering computer vision, deep learning, sensor fusion, embedded systems, and defense product strategy. Growth strategy. Variene builds from its position in the Greek and wider European defense ecosystem, where it participates in multiple national and European defense programs and is positioned to access European defense funding mechanisms. From that base it scales three ways. It integrates its perception layer into platforms operated by primes and partners across Europe and the United States, so reach grows without building hardware. It deepens a proprietary dataset that strengthens every model and raises the barrier to entry. And it extends one common engine across new sensors, mission domains, and dual use markets such as maritime monitoring and critical infrastructure.

Funding Needs

Variene is led by two principals. Stelios Koroneos, founder and CEO, set the company's sovereign autonomy thesis and leads its technology direction. He is an embedded systems engineer with depth in hardware, satellite communications, and wireless networks, alongside senior product roles in maritime data and vessel intelligence. Alex Moukas, President, is a serial entrepreneur and multi-venture operator who founded and led a technology company from inception through a NASDAQ listing and a follow on public offering, and has since built and scaled multiple ventures. They are backed by a team of around twenty across engineering, research, and business development, covering computer vision, deep learning, sensor fusion, embedded systems, and defense product strategy. Growth strategy. Variene builds from its position in the Greek and wider European defense ecosystem, where it participates in multiple national and European defense programs and is positioned to access European defense funding mechanisms. From that base it scales three ways. It integrates its perception layer into platforms operated by primes and partners across Europe and the United States, so reach grows without building hardware. It deepens a proprietary dataset that strengthens every model and raises the barrier to entry. And it extends one common engine across new sensors, mission domains, and dual use markets such as maritime monitoring and critical infrastructure.