Axitea: Converging Security Under Test Between Robots and AI
Keeping physical security and cyber security within a single operational architecture has long been a hallmark of Axitea, which Edge9 observed during a deep dive day hosted by the company, dedicated to showcasing a patrolling solution based on autonomous mobile ground robots. As a Global Security Provider with about 1,000 employees, a clientele of over 18,000 companies, and revenues nearing 100 million euros at the Group level, Axitea started the morning with the robot and ended it at the cyber SOC, passing through statements from the CEO and a visit to the physical security operations center. Four stages that describe as many levels of the same value chain.
A Patrolling Robot for Large Sites
The device presented is an autonomous mobile ground robot designed to patrol large areas where human presence proves inefficient or exposed to risks, from industrial sites to photovoltaic fields. The autonomous navigation is developed by InfoSolution, a technology partner that provides the guiding platform on which Axitea has built its offering. The robot moves by combining a 16-plane LIDAR, which reconstructs the surrounding environment in 3D and allows for avoiding unmapped obstacles, and GNSS satellite localization with GPS, Galileo, BeiDou, and GLONASS for outdoor contexts over large surfaces. A visual odometry camera helps estimate movement when laser references are scarce. All calculations are onboard, on two NVIDIA platforms: one dedicated to navigation, the other to higher-level functions, like intruder detection via neural network and communication with the SOC. Connectivity is redundant on 4G, 5G, and Wi-Fi with dual SIM, but the robot can complete its mission even if the network fails.
Patrolling occurs via waypoints defined on a map created during installation, with the possibility to plan scheduled or random missions. Specific actions can be associated with each waypoint, such as turning to frame a gate, taking a snapshot, checking for the presence of a fire extinguisher, or a closed door. The base sensor suite relies on optical and thermal cameras with 360-degree vision, but the same robotic base can host additional sensors, from airborne substance detection to thermal cameras for fire prevention. The range includes four models with load capacities from 30 kilograms up to 500 kilograms, allowing the vehicle to adapt to the operational environment. The IP66 certification allows for use in any weather conditions.
Integration with Axitea's operations center allows remote operators to receive snapshots of intrusions, take control of the robot via joystick, activate deterrent lights, and broadcast pre-recorded audio messages. The first pilot project is the Milan sports center in Milanello, where the robot patrols a perimeter that combines lawn, walkways, and forested areas. The choice of the pilot project clarifies the company's declared positioning: the robot does not replace security personnel but covers intervals where human presence would be difficult or ineffective, such as night patrols in expansive or dangerous areas. The solution also relies on iVideoHub, the video analysis platform with artificial intelligence that Axitea already uses at fixed sites.
The Group's Evolution and Market Analysis
The inclusion of the robot in the service catalog is one of the signals of an expansive phase for Axitea. In May 2025, private equity fund Argos entered the company's capital, and in February 2026, the newly formed Axitea Group announced its first consolidation operation with the acquisition of Surveye, a system integrator active for thirty years in surveillance systems, intrusion detection, access control, and fire detection. Surveye retains its CEO and identity, while the total Group perimeter reaches 1,000 employees, 18,500 customers, and revenues around 100 million euros. The operation is described as the first step in a buy-and-build strategy aimed at strengthening system integration and building a converging security model, both physical and cyber, based on data and artificial intelligence.
Marco Bavazzano, CEO of Axitea, reads the moment with concrete concern for the macroeconomic framework, but also with substantial demand resilience, because security, he reminds, is a primary need that has shown to endure even in previous crises, from the pandemic onwards. The most interesting part of his analysis concerns cybersecurity, where the manager identifies a shift in scenario not yet absorbed by Italian companies, namely the multiplication of non-human identities. The attack surface, he notes, has expanded enormously due to the proliferation of these identities. Multi-agent architectures, integrating with information systems and business applications, generate new potential access points that are often overlooked. While awareness of risks from data moving to public models is now widespread at least in principle, regarding the increased attack surface driven by the rapid adoption of AI agents, there is, according to Bavazzano, a gap to be filled.
The other area of focus is the Italian manufacturing sector. Bavazzano identifies four pillars upon which the competitiveness of companies in the sector depends: asset security, performance, quality control, and worker safety. On all these issues, he claims, computer vision and AI can offer even customized solutions, but most manufacturing companies still do not know how to coherently seize them. Industry 4.0 and 5.0 are seen as useful but incomplete tools, as they have pushed for the replacement of obsolete machinery without fully integrating production data with analytical platforms capable of translating into real governance. Hence the invitation to rethink future incentives in a less bureaucratic way, more oriented toward actual digital evolution.
An Operations Center That Correlates Alarms and Images
The operational translation of this vision is seen in the physical security operations center, which currently manages about 110,000 alarms per month from customer systems. Over the past two years, Axitea has developed a proprietary application layer above hardware vendor platforms, aiming to standardize operator work and feed it with process metrics comparable across different clients. The qualitative leap concerns the correlation between alarm sensors and cameras: rather than generating a generic alarm that pushes the operator to check the entire array of cameras on a site, the system links each sensor to the corresponding video channel. When a siren goes off on the right side of a Smart Tower, for example, the operator immediately sees only that channel and can launch recording in conjunction with the event.
The effect on the workflow is significant. Of the roughly 110,000 monthly alarms for physical security, about 20,000 are automatically managed by the new platform, while 90,000 still pass under the watchful eyes of operators, manned by a team of 70 professionals distributed across two locations. The ability to qualify alarms also changes the propensity for field intervention: for the so-called qualified alarms, where the system provides a precise context, patrols are dispatched in about three cases out of 800. When the alarm is blind and it is unclear what is happening, the intervention rate rises between 60% and 70%. The difference measures how much the sensor-camera correlation contributes to mitigating the