With MarsLume AI, ZKTeco has taken a decisive step beyond biometrics-based access control, positioning MarsLume as an AI Cognitive Space Platform at the center of ZKTeco systems. With the latest integration of new features, most notably natural-language controls and agentic AI, ZKTeco is cementing the role of MarsLume AI as the central intelligence layer that enables intuitive usage and automated processes in video security and beyond.
From access control to spatial intelligence
For ZKTeco, the positioning of MarsLume AI as an AI Cognitive Space Platform represents more than just the addition of new, user-facing features. It signals a shift in how ZKTeco architectures interact with the spaces the systems are meant to secure—from identity-based access control to scenario-aware management, integrating access control video security and other functions such as alarms.
Access control has traditionally answered a few narrowly defined questions: Who is allowed to enter, when and where? Access control systems that integrated surveillance cameras used them as auxiliary tools to secure entry points more efficiently. With MarsLume AI, however, ZKTeco architectures are enabled to understand what is happening in the physical environment before, after and during the access event, as well as fully independently.
For ZKTeco users, this means the existing biometric identity, credential, access event and attendance data become part of a broader, AI-enabled operational intelligence layer. Instead of treating access control, video security, alarms and security event reporting as different domains that are connected with each other on the surface level, MarsLume AI enables unified workflows that overcome system boundaries.
A few examples: Entrance events become legible in the context of video footage and broad data analytics. Credentials used at the door add depth to the analysis of the movement of persons across the site. A safety violation—for example a worker not wearing the required protective equipment—can trigger both visual verification and an access control response.
In short, MarsLume AI enables ZKTeco users to understand and manage comprehensively the entire space of their site instead of only controlling access to it.
Natural language enables intention-based collaboration
At the center of MarsLume AI’s new capabilities is natural-language interaction—a shift that changes the relationship between security teams, the space they manage and the wider organization they belong to. Traditionally, security system interfaces required them to navigate menus, filters, timelines, rules engines, and technical configuration logic, often creating a gap between the human understanding of the scene and the understanding of the technical system. MarsLume AI bridges the gap by enabling users to communicate with the system in a scenario-driven way that is natural to them.
In practice, instead of manually sifting through hours of disconnected footage from multiple cameras, users can describe a person of interest or event in plain language—“find the person wearing a red jacket who exited through the lobby this morning,” for example—and receive results within seconds.
Operating systems before MarsLume and with MarsLume will feel significantly different: Previously, investigating an incident took hours and was prone to important pieces of evidence being lost due to human error. Reconstructing the course of events based on cross-referencing disconnected video timelines and access logs. Now, MarsLume AI understands what is happening across the premise, connecting data from different domains. Incident investigations only require the operator to give the system a prompt, and MarsLume shares its understanding of the scene within seconds at highest accuracy.
Alarm setup is just as easy: Instead of configuring complex rules, users can describe their intention in a sentence, and MarsLume AI transforms it into an instantly workable detection rule without requiring training data or technical setup. The result is not just a system that is more convenient to use. Natural language is the first of multiple decisive ways in which MarsLume changes the relationship between human operators and security systems—from command-based operation to intention-based collaboration.
Agentic AI: From detecting events to taking autonomous action
MarsLume AI also pushes the boundary with deep integration of AI Agents—enabling fully automated workflows that move beyond detecting events and alerting operators, to taking action autonomously. When, for example, a security-relevant event is being detected by the system, the AI Agent can locate and analyze all relevant access events and video footage of the persons involved and trigger alert escalation, from voice broadcast to access-permission restrictions within seconds of the event, as well as automated incident reports to supplant the understanding of human operators and compliance reasons.
The orchestration between operators and the AI Agent is governed by configurable rules, permissions and approval logic. Users can define what actions the system can take autonomously and which ones need human-in-the-loop confirmation, as well as which users may modify workflows. Customization options also enable users to address operational and regulatory requirements—from daily security reports to EHS compliance and management briefings. The result is an AI-enabled system that executes repetitive and time-sensitive tasks automatically, without diffusing human responsibility and oversight.
Here, too, operating systems before MarsLume and with MarsLume will feel significantly different. Think of the scenario of a PPE violation on the factory floor—something that went undetected in most cases without the instantaneous understanding of the AI Cognitive Space Platform and its agentic intervention: When MarsLume AI detects a worker operating machinery without the required protection, the system does not simply log the event. The AI Agent pulls the worker's access records, retrieves footage from the relevant cameras, triggers a voice broadcast at the location, temporarily restricts the worker's access permissions until compliance is confirmed, and generates an incident report with video evidence, identity data and a timestamp — dispatched automatically to the EHS manager and floor supervisor.
The underlying conceptual shift
With MarsLume AI, ZKTeco systems become proactive partners that enable users to secure sites more efficiently and unlock the systems’ potentials beyond security. Previously, security systems were largely passive: They recorded and stored video footage and access data, and waited to be queried. With MarsLume AI, security systems are enabled to become conceptually active—capable of understanding context, generating insight and initiating action without waiting for the operator to ask the right question. The term AI Cognitive Space Platform is more than a new framing for broader AI integration into security systems: It describes the creation of AI-human collaborative understanding of space of a new level.
Benefits for users, organizations and integrators
For end users and security teams, the most immediate benefit is a reduction in the operational burden of managing complex, multi-domain systems. Natural-language search and alarm setup along with automated reporting reduce the time spent on forensic investigation and documentation, freeing capacities for higher-level tasks.
For organizations, MarsLume AI offers an upgrade path that protects previous investments. Existing cameras, access control devices and other ZKTeco solutions can be integrated with MarsLume AI, adding a smart cognitive layer to established systems and workflows. Organizations can begin with selected high-value tasks and scenarios—perimeter protection, safety compliance detection, visitor tracking—and expand gradually as operational needs evolve.
For integrators, MarsLume AI’s support for RTSP, ONVIF and GB28181 means it can be deployed flexibly in systems integrating cameras from multiple vendors, broadening the range of projects for which it is suitable. Its modular architecture and plug-and-play compatibility with existing ZKTeco technologies and platforms such as BioCV multimodal AI and the ZKBio CVSecurity platform enable integrators to position MarsLume AI as a value-adding layer in both new deployments and existing infrastructures.
Final thoughts: Enhancing ZKTeco’s foundation in access control with MarsLume AI
ZKTeco has ever since its inception designed solutions that put identity in the center of security operations, with biometrics-based systems that enable operators to know at anytime who is entering and exiting their site. MarsLume AI does not replace this foundation; instead, it enhances it significantly. With MarsLume AI, operators do not only know who is entering and leaving the protected space, it also gives them AI-enhanced capacity to act based on full situational awareness about what is happening at any time in every corner of their premises.
For more information about ZKTeco products and solutions, please visit
www.zkteco.com