ZKTeco becomes full-scale enabler of intelligent ecosystems with MarsLume AI and RS3

Technological breakthroughs always open new doors for companies to introduce new products. With MarsLume AI and RS3, ZKTeco is taking this well-known logic a step further. Boosted by breakthroughs in multimodal AI, amid robust customer demand and accelerating adoption, MarsLume AI and RS3 represent a fundamental shift in ZKTeco’s philosophy—taking a big step far beyond the biometric-based access control solutions that the company has been known for for decades.
 
MarsLume AI is a cognitive platform built on multimodal large models and intelligent agents. Unlike traditional security analytics that rely on rigid rules, it provides a "semantic understanding" of the scenes at hand—for example when a person uses an access control terminal and their credentials are captured by a reader while a security camera observes the scene.
 
RS3 complements MarsLume AI as a flexible and adaptable orchestration platform that can serve as the "cognitive backbone" of a whole ecosystem of smart devices. As a software-based solution, RS3 can coordinate all components of the system, ensuring seamless communication and intelligent, automated responses. In real-life scenarios, it is the link that allows MarsLume AI's cognitive capabilities to be integrated with existing access control, video and IoT infrastructures.
 

Access control becomes a decision-making node

Even though MarsLume AI and RS3 mark a decisive step beyond ZKTeco’s traditional field of expertise, access control will continue to be a cornerstone of the company’s expanding offering. With ZKTeco, access control becomes more than just a system that decides whether to let somebody in or not, depending on their credentials.
 

Understood in a new, wider sense, as a decision-making node in an intelligent ecosystem, access control devices powered by MarsLume AI produce structured, high-value data: The identity of each person passing through, the time of their passage, permission status, and potentially much more.
 
This can make access control logs richer than most security video footage—enabling ZKTeco to make full use of the data beyond their original realm.
 
Combined with the unstructured data generated by security video streams, a person in a red jacket captured while walking through the hallways of a building, for example, becomes an identified user with a history of entering and leaving the building, as well as a multitude of past interactions with its IoT ecosystem.
 

Fusion understanding

One key feature of MarsLume AI is the semantic comprehension of scenes, powered by the fusion of visual data and language models.
 
On the user interface side, this means security personnel can give the system natural-language prompts, such as for example: “Check whether the person in the red jacket previously passed the gate carrying a large package out of the building.” MarsLume AI doesn’t just scan for “man”, “red jacket” and “package”—it also understands contexts and sequences, enabling security personnel to find the correct footage within seconds and match the results with user identities.
 
Another key feature is semantic monitoring. MarsLume AI moves beyond rule-based alerts (e.g., "motion detected in restricted area") as it understands behaviors and contextual interactions. This allows security personnel to define alerts using descriptive language, such as "send a notification when somebody carries a large package out of the building.” The results are narrowly defined, readily applicable rules, leading to fewer incidents missed and falsely triggered alarms.
 

Synergies between platforms

At the heart of ZKTeco's new ecosystem is a synergistic relationship between three platforms. Aside from MarsLume AI (“the cognitive brain”) and RS3 (the "cognitive backbone"), ZKBio CVSecurity acts as the operational layer, managing workflows, devices and providing the unified user interface for all interactions.
 
In a typical scenario—such as, for example, a tailgating attempt at the entrance to a restricted building—it would look like this:
 
The access control terminal, together with other edge devices such as cameras, detect the anomaly in real time—powered by MarsLume AI. At the platform level, RS3 confirms the incident, matches the identities of those involved against all available databases, analyses movement pattern and correlates the data with access control logs.
 
At the user interface level, ZKBio CVSecurity—the access control management suite that ZKTeco users are already familiar with—alerts pop up for security teams to see and pre-defined sequences of actions are triggered, such as locking doors or sounding the building alarm.
 

Flexible deployment

One thing that hasn’t changed about ZKTeco’s philosophy is the focus on openness and easy integration. The new platforms, too, are designed for flexibility and scalability.
 
RS3 supports on-premise deployment on local servers, ensuring data sovereignty and compliance, as well as cloud hosting, for organizations seeking easy integration with their existing infrastructures and instant access to the new features.
 
Most large-scale deployments, however, adopt a hybrid model, with local servers on-site combined with cloud services for centralized analysis and management. This guarantees real-time responsiveness and scalability across different sites, making MarsLume AI and RS3 suitable for smart cities, industrial campuses and even nationwide deployments.
 
Users who already have ZKTeco architectures in place can gain access to the new features via a simple cloud upgrade, without requiring new hardware. Alternatively, they can add MarsLume-powered and RS3-compatible servers, turning them into to “cognitive brain” of their infrastructures.
 
ZKTeco designed the integration points of MarsLume AI, RS3 and ZKBio CVSecurity with the goal to enhance the user experience—without disrupting their workflows. ZKBio CVSecurity’s video search box becomes a natural language input field, supporting written and spoken prompts to ensure fast and flexible workflows. New modules, such as multimodal search, trajectory mapping and crowd analytics, are integrated as extension of the existing interface. With MarsLume AI and RS3, ZKBio CVSecurity also adds automated incident response options, guiding the workflows of security teams and making their tasks more intuitive.
 
Meanwhile, MarsLume AI and RS3 can also be integrated with third-party enterprise systems, from human resources to resource planning, manufacturing execution, building management and IoT—enabling the ZKTeco platforms to host and extend third-party AI. This also includes AI agents, such as Coze or Dify.
 

Conclusion: Designing the ecosystems of the future

ZKTeco sees the move beyond access control as a new foundation to build new solutions and technologies upon. It would indeed be surprising if the ambition to build intelligent ecosystems ends here.
 
Instead, ZKTeco seeks to become the “global leader in intelligent space evolution services.” It has therefore identified four strategic domains in which AI will have an ever-bigger impact, and ZKTeco seeks to play an ever-bigger role in them:
 
In smart spaces, security systems will become platforms that optimize energy and resource use, in line with organizations’ sustainability goals. In smart offices, advances in AI will unlock more productivity and efficiency through workforce and time management. In the field of digital identity, blockchain-integrated biometrics will build the trust fabric of the digital world. In smart commerce, AI will drive retail and customer engagement, enhancing consumer experience and business growth.
 
These are big visions, but with MarsLume AI and RS3, ZKTeco has already demonstrated that it can follow through on big, visionary ideas and turn them into market-ready solutions.
For more information about ZKTeco products and solutions, please visit http://www.zkteco.com.