AI agents are becoming increasingly versatile, seamlessly interacting with humans through text and speech. As organizations strategize the integration of generative AI into their operations, IT leaders should closely monitor the rise of a breakthrough paradigm: multiagent systems. These systems epitomize the next frontier in automation and artificial intelligence.
Multiagent systems consist of several agents working collaboratively to execute tasks aimed at a common objective. These objectives could range from automating payroll and human resources processes to advancing software development, leveraging text, images, audio, and video inputs from sophisticated large language models (LLMs).
According to a Capgemini survey, a staggering 82% of business leaders anticipate incorporating multiagent systems into their operations within the next one to three years. These agents promise to streamline a wide variety of tasks, from drafting emails and software code to data analysis, transforming the landscape of workplace productivity.
The operational intricacies of multiagent systems differ depending on the tasks and goals for which they are designed.
All Aboard the Multiagent Train
Think of multiagent systems as akin to conductors orchestrating a train. You’ll find a lead conductor—the “boss” agent—who delegates tasks to a range of subagents. A human user initiates interaction with the lead conductor via a traditional user interface, like an LLM prompt window, which then triggers a series of actions handled by the subagents.
These agents collaborate with other digital tools, systems, and humans, continually accessing corporate databases to enrich their organizational knowledge. Notably, they learn from their task history, human feedback, and other data inputs, allowing them to continually enhance their performance and adapt to changing environments.
In essence, multiagent systems are self-regulating and iterative, mirroring human workers in their ability to improve and adapt over time. In a landscape where organizations are leveraging generative AI to generate value, multiagents could very well be the key to unlocking unprecedented levels of operational productivity.
The value these systems promise stems from their ability to automate complex tasks characterized by highly variable inputs and outputs, scenarios that have traditionally posed significant challenges to automation.
Making Automation Actionable
A compelling example showcased by McKinsey is the automation of business trip bookings. Consider the myriad details required to organize travel, accommodation, dining, and more. While some aspects have been automated to a degree, the required variability in inputs and outputs has historically been too challenging and costly to automate fully.
Envision a multiagent system providing “actionable automation” across departments like sales and marketing, human resources, and IT operations. An agent could generate a sales analysis report, collaborating with other agents to gather relevant sales data, draft the document, ensure compliance with corporate standards, and refine the report accordingly. McKinsey also identifies loan underwriting, code modernization, and marketing collateral as potential areas ripe for multiagent integration.
However, the potential for multiagent systems extends beyond the digital space. These agents could manage electrical systems, from elevators to HVAC systems, ensuring optimal temperatures and lighting across various zones—areas already benefiting from a high degree of automation.
While this largely remains speculative, the breadth of potential use cases depends on specific business requirements. Automating these tasks would enable human employees to focus on more nuanced aspects of the business, such as fostering collaboration and enhancing customer engagement, ultimately boosting both employee and customer satisfaction scores.
However, a significant challenge lies in maintaining digital resiliency in multiagent systems. The question remains: if one agent fails, does the entire system collapse? The tech industry has encountered similar issues with robotic process automation, where rule-based bots falter in the face of variability. Fully autonomous agents must be capable of self-correction to meet their objectives.
Until such advancements are realized, having a human-in-the-loop to initiate kill switches or perform rollbacks is crucial as organizations experiment with these systems.
Preparing Your Organization for the Multiagent Revolution
The ultimate goal is to ensure that multiagent systems align with organizational objectives to deliver the desired business outcomes. As an IT leader, you must be prepared to support these systems if your organization chooses to adopt them.
This preparation involves front-line coders, DevOps practitioners, and hardware engineers—all must be ready for dynamic change, whether incorporating single digital assistants or vast arrays of autonomous agents. A modular approach to system architecture—one that simplifies development, testing, and troubleshooting—can minimize disruptions. Similarly, optimizing the infrastructure is essential.
Dell Technologies’ Dell AI Factory offers a comprehensive solution, integrating AI innovation, services, and an extensive ecosystem of partners to help organizations achieve their AI goals. Dell’s professional services team assists organizations in preparing their data and identifying and executing relevant use cases.
Learn more about the Dell AI Factory.