How SoftServe used AGNTCY to implement a multi-agent intelligence system for video monitoring

Traditional video monitoring systems face a fundamental limitation. They’re built as single, massive applications that struggle to adapt. When businesses need to add new capabilities such as facial recognition, crowd density analysis, or anomaly detection, they often require complete system overhauls.

AI video monitoring requires an architecture that can handle multiple tasks simultaneously, adapt to changing needs, and manage large volumes of video data in real time, all while prioritizing security of the system. As a result, despite the benefits of this technology, organizations may struggle to build and scale such systems. 

SoftServe, an IT consulting and digital services provider, tackled some of the biggest challenges in AI-powered video monitoring by using the components developed by the AGNTCY, an open source collective for inter-agent collaboration. By distributing intelligence across independent, task-specific agents, SoftServe achieved real-time adaptability and scalability without system disruptions. 

In a recent whitepaper, SoftServe explains how they transformed traditional video monitoring from rigid, monolithic systems into a dynamic, collaborative multi-agent system. By using the Internet of Agents architecture through AGNTCY’s open source framework, SoftServe created solutions to overcome key architectural challenges and implement a multi-agent intelligence system for video monitoring. For a comprehensive overview and technical details, explore the whitepaper.

Challenge 1: Orchestrating multi-agent systems

In agentic systems, intelligence can be distributed across multiple agents. For example, individual agents can be responsible for a specific task, such as object detection, anomaly recognition, or data retrieval. However, without a clear orchestration framework, tasks can be misaligned, leading to inefficiencies or system failures. Dependencies between agents can also introduce scheduling and error handling. 

Solution: AGNTCY simplifies orchestration by introducing a centralized coordination layer so agents can communicate and collaborate effectively. Through SLIM (Secure Low-latency Interactive Messaging) AGNTCY enables asynchronous, low-latency communication between agents. This significantly reduces complexity of agent orchestration and ensures real-time responsiveness and smoother task execution. 

Challenge 2: Integration complexity 

With multiple agents, each agent can have its own data format and communication protocol. A lack of standardization in inter-agent communication creates integration bottlenecks and makes it difficult to introduce new agents or replace existing ones. 

Solution: With a plug-and-play modular architecture, agents are described using Open Agent Specification Format (OASF) JSON manifests. These manifests define each agent’s inputs, outputs, and communication parameters, ensuring consistent integration across the system. This design streamlines integration and ensures agents can be switched, reused, or added without disturbing the overall system. Also, SoftServe leveraged ACP (Agent Connect Protocol) from AGNTCY, a universal SDK (Software Development Kit) that enables agents built on different frameworks to communicate effortlessly. This allowed SoftServe to use the best tool for each job without worrying about integration complexity.

Challenge 3: Scaling for real-time performance

Processing live video streams in real-time requires systems that can handle high throughput and low latency. Traditional architectures often struggle to scale when faced with increasing data volumes. Tightly coupled architectures can constrict systems as data volumes grow, while overloading a single agent or node can degrade system performance.

Solution: AGNTCY’s decentralized architecture supports independent operation of agents, ensuring that no single component becomes an obstruction. With SLIM, the system can handle high data volumes and remain responsive under heavy workloads. It allowed SoftServe to maintain real-time performance with their video streams. Manufacturing clients reported 40% faster incident detection compared to legacy systems.

Challenge 4: Modularization and maintenance

With tightly coupled architectures, it can be a challenge to update or replace components without causing major headaches. Even minor updates to one agent can lead to downtime and increased costs. 

Solution: AGNTCY’s modular design decouples agents so they can function independently. Each agent handles specific tasks, dramatically simplifying debugging and system maintenance. When issues arise, SoftServe can isolate the problem and update individual agents without impacting the rest of their system. This allows agents to improve over time. 

Explore more real-world applications with AGNTCY

SoftServe’s success demonstrates that the Internet of Agents isn’t a distant future concept, but rather a practical architecture delivering value today. As multi-agent systems become more relevant across industries, AGNTCY’s framework provides a scalable and adaptable solution. Download the full whitepaper to explore additional real-world applications, lessons learned, and a deeper dive into the implementation details.