Follow ZDNET: Add us as a favorite source On Google.
ZDNET Highlights
- The number of agents is continuously increasing, increasing the risk of spread.
- Professionals should consider using agent management systems.
- These systems can help manage the spread of agents, but be aware of the challenges.
Enterprises have 28.6 million active agents worldwide, an estimated over 2.2 billion According to Statista, by 2030.
Also: These top 30 AI agents offer a mix of tasks and autonomy
Agent wranglers need to bring management sensitivities to this growing sector. So, can AI agent proliferation be controlled? Some vendors are trying this, giving rise to a new technology category, agent management systems, which are tasked with managing networks of AI agents.
build a platform
An agent management platform essentially acts as a digital human resources department for AI agents, and experts suggest now is the right time for such offerings.
Agents running outside the management framework are essentially the AI equivalent of shadow IT.
“It works until it stops working, and when it stops working, you have no audit trail, no version control, and no governance.” noted Shelley Palmer, professor at Syracuse University and CEO of The Palmer Group.
Agent management solutions on the market include Google Vertex AI Agent Builder, Amazon Bedrock Agents, Microsoft 365 Copilot, Decagon AI, and Sierra AI, which serve a variety of purposes from orchestrating systems to multi-agent automation.
Also: 5 Myths of the Agentic Coding Apocalypse
These platforms are essential to the future of agentic automation. The key to success is “treating agents as infrastructure rather than features,” said Deeptmoy Sanyal, principal engineer at CrowdStrike.
Agents are not one-time creations. “The problem is that you end up with dozens of agents with no shared reference model, no consistent governance, and no reusable patterns,” Sanyal said. “A proper management platform gives you composable primitives, multi-tenant isolation, model routing across LLM providers, and an overview of what agents are actually doing.”
fight the spread
As agents are growing by the millions and handling everything from sales to software development, the major hurdle is that they all want access to the same data.
“This creates an AI governance challenge,” said Manu Narayan, CIO of GitLab. “If you don’t intentionally build your AI stack, you can end up with dozens of vendors and all their agents holding the keys to the empire.”
Also: How to build better AI agents for your business – without creating trust issues
This situation leads to the proliferation of agents, “a fragmented ecosystem of loosely managed agents with inconsistent behavior, duplicated functionality, and unclear ownership,” said Yash Vijay Patil, a software engineer at Texas A&M University. “Without strong governance, this spread could lead to operational inefficiencies and increased risk exposure.”
Many vendors and internal teams are building agent solutions for specific use cases, but they often lack a shared identity model, lifecycle policies or risk framework, said Monica Malik, principal data and AI engineer at AT&T. “That approach creates duplication, inconsistent behavior, hidden costs, and security risks. The problem will not be too few agents, but too many unmanaged agents.”
Then, the popularity of consumer alternatives like OpenClave has increased the complexity of agent networks, said Brian Jackson, principal research director at Info-Tech Research Group. “It’s safe to assume that some employees will try to automate their work tasks with them. This creates problems for tracking all the agents you have deployed in an enterprise environment. While various management platforms claim they can discover the agents deployed in your systems, the truth is that they are limited by the identity management layer.”
Agent management platforms offer benefits like overview, Jackson said, so you know which agents you’re using and what they’re doing.
Also: Give your ‘human-level agents’ a head start with these 3 best practices
Additionally, these platforms enable “using a central policy to set guardrails for what agents can and cannot do and keep them aligned with enterprise goals.” Ultimately, these systems enable value capture, as they “monitor performance over time and ensure that agents’ costs and outputs are in line with expectations, and add value to the work,” he said.
The role of such management platforms, AT&T’s Malik said, is to “provide a control layer for how organizations deploy, monitor, secure and improve their agents over time.” “The key benefit of these platforms is not just orchestration, but operational discipline: visibility into what agents are doing; where they’re pulling data from; how they’re making decisions; when human oversight is needed.”
Understanding Market Trends
However, Jackson said competition among vendors to capture the agents’ management space is fierce.
“This will be a strategic position where enterprises are building their own workflows and building deeper relationships across an ecosystem,” he said.
Plus: 5 security strategies your business can’t get wrong in the age of AI – and why they’re important
As a result, multiple agent implementations will be tied to familiar systems of record within different areas of the business, Jackson added. “You get into a situation where marketing is managing agents from a CRM platform, while IT is managing agents from an asset management and overview platform.”
As agents become more autonomous, “defining clear boundaries, monitoring behavior, and maintaining trust will be important,” said Texas A&M’s Patil. “In short, agent management platforms provide powerful leverage, but only when paired with disciplined governance and thoughtful adoption strategies.”
“It’s a challenge to overcome complexity when agents operate on multiple interconnected systems simultaneously,” Narayan said.
“Integration through agent management platforms helps,” he said. “They establish context, permission models, security controls, and data boundaries that simplify agent orchestration at scale. Combining this type of platform with a hub-and-spoke model can help you become more intentional in your AI stack without slowing the pace of adoption.”
applying technology
Another challenge with agent management platforms, Malik said, is that “they are harder to change than most cloud alternatives because they shape workflows, integrations, permissions, and operating models.”
This is why adopting agents should be an enterprise decision. All stakeholder departments – from engineering to security, legal, data governance to business owners – need to be involved in decisions about the agent management platform. “The primary hurdle is avoiding fragmented adoption. Organizations must view agent platforms as long-term operational infrastructure, not just another purchase of AI equipment,” Malik said.
Agent platform decisions are difficult to reverse because they are deeply embedded in workflows, data pipelines and business logic, Patil said. “Evaluate platforms based on interoperability, extensibility, vendor lock-in risk, and support for open standards. Importantly, the decision should not be left to engineering alone – cross-functional stakeholders should be involved, including security, data, and business leaders.”
Also: Why enterprise AI agents could become the ultimate insider threat
Additionally, professionals should remember that “getting data and workflows out of legacy software platforms is already difficult,” Jackson said. “Adding an AI layer on top of that means the integration goes even deeper into the platform. Trying to move an agent management system would be like trying to do a brain transplant.”
Therefore, businesses should prioritize flexibility when moving to an agent management platform. “Evaluating where you’re comfortable placing bets on the platform versus trying to place bets on a self-hosted platform,” Jackson said.
“Given the unpredictability of consumption costs for agent workloads, it may be wise to design a system that leverages internal infrastructure and avoids tying business processes to metered fees or consumption-based pricing.”
Professionals should treat the development and implementation of an agent management platform “like a database selection, not like a SaaS tool evaluation,” Sanyal said. “Involve platform engineering, security, and legal from day one. Not after the pilot is successful. Also, the decision should not rest with a single business owner. It requires platform engineering, security, and who owns your identity and access model to be in the same room.”
