Today's digital ecosystem demands that individuals and businesses operate across multiple devices: a smartphone to capture information, a desktop to process it, and an IoT device to execute an action. However, artificial intelligence agents capable of handling such collaborative tasks across devices remain a major technical challenge. The recently introduced DevicesWorld benchmark highlights this gap by evaluating LLM-based agents in an environment that integrates mobile, desktop, and IoT. The results are revealing: even the most advanced systems achieve only 12.5% success on tasks requiring coordination of information and actions across different environments. This failure is not anecdotal; it points to a critical gap in enterprise automation.
DevicesWorld consists of over six thousand tasks designed to simulate real user needs involving multiple devices. Each task includes a natural-language goal, participating devices, initial states, executable actions, and automatic verifiers. The evaluated agents showed recurring problems: they get stuck acquiring information, confuse source and output devices, or terminate before meeting all conditions. About 28.7% of failures met at least one scoring criterion, indicating partial progress but not final integration. This reflects the complexity of orchestrating workflows that cross hardware and software boundaries.
From a technical perspective, the difficulty lies in heterogeneous interfaces, state synchronization, and device dependency management. An agent operating on a mobile device cannot assume information is available on the desktop without a transfer and validation mechanism. Additionally, IoT environments introduce latency and proprietary protocols. For businesses, this means automating processes that involve multiple platforms requires a much more robust software development approach than a monolithic agent. This is where the need for custom software applications comes into play, integrating device communication, state persistence, and action orchestration natively.
The cloud becomes the natural enabler to overcome these limitations. Services like AWS and Azure provide infrastructure to synchronize data across devices, run agents in serverless environments, and maintain a centralized coordination layer. Q2BSTUDIO provides cloud services that allow building scalable architectures capable of managing multi-device complexity. For example, an agent can capture an image from a mobile device, upload it to an S3 bucket, have it processed by an AWS Lambda function, and send the result to a desktop dashboard, all orchestrated via events. This cloud integration not only facilitates communication but also provides traceability and security.
Cybersecurity is another fundamental pillar when dealing with multiple devices. Each entry point — mobile, desktop, IoT — represents a potential attack surface. Agents handling sensitive information must ensure authentication, encryption, and data integrity in transit and at rest. Businesses need cybersecurity solutions that include penetration testing in multi-device environments and granular access policies. Q2BSTUDIO integrates these practices into its developments, ensuring every agent and communication channel is protected against intrusions.
The data generated by multi-device agents is a goldmine for business intelligence. Tools like Power BI enable visualizing agent performance, detecting bottlenecks, and optimizing workflows. For instance, if an agent systematically fails to transfer data from mobile to desktop, a BI dashboard can alert on that pattern. Q2BSTUDIO offers Business Intelligence services that transform this data into actionable insights, helping companies improve efficiency of their automated processes.
The artificial intelligence agents themselves are at the center of this transformation. DevicesWorld results show that current LLMs are not sufficient for multi-device tasks; an additional layer of reasoning and planning is needed. Q2BSTUDIO develops custom AI agents that incorporate memory modules, context management, and the ability to interact with APIs from different devices. These agents not only understand natural language but also execute actions in heterogeneous environments, such as opening a desktop application, retrieving data from an IoT sensor, and sending a summary to the user's mobile phone.
Process automation is the ultimate goal. Integrating AI agents with workflows that cross devices allows companies to reduce manual intervention, increase response speed, and improve accuracy. Q2BSTUDIO offers process automation through software that connects applications, devices, and cloud services. A typical case would be incident management in an industrial plant: an IoT sensor detects an anomaly, a mobile agent notifies the technician, the technician processes the information on their desktop, and the system automatically updates inventory in the cloud. This chain of actions requires the kind of coordination that DevicesWorld tests.
In conclusion, DevicesWorld exposes an uncomfortable reality: AI agents are not yet ready for the real multi-device world. However, it also opens an opportunity for companies to invest in robust architectures, custom software, and cloud services that overcome these barriers. Q2BSTUDIO, with its expertise in cross-platform application development, cloud, cybersecurity, BI, and AI, is well positioned to help organizations build systems that not only execute tasks but truly collaborate across devices. This is the next step in the evolution of enterprise automation.
