AI Research Forum and Summit Focused on Agentic AI Announced
A New Platform to Explore the Future of Autonomous, Goal-Driven AI Systems.
As AI systems gain autonomy and begin to operate with less direct human oversight, a new category of AI tools is emerging to manage the risks: guardian agents. These technologies are designed to monitor, guide, and, when necessary, intervene in the behavior of other AI agents—particularly in enterprise settings where the stakes are high.
According to Gartner Inc., guardian agents will represent 10 to 15% of the agentic AI market by 2030, signaling their growing importance in AI governance and cybersecurity strategies.
Guardian agents operate both as AI assistants that support tasks such as content review and monitoring, and as autonomous or semi-autonomous systems capable of executing or blocking actions based on predefined goals.
A recent Gartner webinar poll revealed that adoption of AI agents is already underway. Among 147 CIOs and IT leaders surveyed, 24% reported deploying a few AI agents (fewer than a dozen), and 4% said they had deployed more than a dozen. Meanwhile, 50% are in the research or experimental phase, and another 17% plan to implement the technology by the end of 2026.
“Agentic AI will lead to unwanted outcomes if it is not controlled with the right guardrails,” said Avivah Litan, VP Distinguished Analyst at Gartner. “Guardian agents leverage a broad spectrum of agentic AI capabilities and AI-based, deterministic evaluations to oversee and manage the full range of agent capabilities, balancing runtime decision making with risk management.”
As enterprise use of agentic AI grows, so does the associated risk. 52% of 125 respondents from the same poll indicated their AI agents focus on internal administrative functions such as IT, HR, and accounting. Meanwhile, 23% reported external, customer-facing applications.
Gartner highlights several major threats to agentic AI systems, including data poisoning, credential hijacking, and agents interacting with malicious or deceptive online sources. These vulnerabilities can lead to unauthorized access, operational disruptions, and reputational harm.
“The rapid acceleration and increasing agency of AI agents necessitates a shift beyond traditional human oversight,” Litan said. “As enterprises move towards complex multi-agent systems that communicate at breakneck speed, humans cannot keep up with the potential for errors and malicious activities. This escalating threat landscape underscores the urgent need for guardian agents, which provide automated oversight, control, and security for AI applications and agents.”
To address these concerns, Gartner recommends CIOs and security leaders prioritize three core functions of guardian agents:
Regardless of how they are deployed, Gartner notes that guardian agents are essential for maintaining the integrity of increasingly complex AI ecosystems. The firm predicts that by 2028, 70% of AI applications will incorporate multi-agent systems—making automated governance tools not just useful, but critical.
A New Platform to Explore the Future of Autonomous, Goal-Driven AI Systems.
The initiative supports organizations in deploying AI-powered agents that boost operational efficiency, drive growth, and reimagine work.
Among 147 CIOs and IT leaders surveyed, 24% reported deploying a few AI agents and 4% said they...
Copyright © 2025. MIT Sloan Management Review Middle East. All rights reserved.
![]() |
Thank you for Signing Up |
Jithin George designs and builds agentic AI systems that operate with autonomy, context awareness, and real-time decision-making capabilities.
His work focuses on creating AI agents that can reason, act, and adapt across complex environments—bridging the gap between intent and execution in enterprise and developer workflows.
Munjal Shah is the co-founder and CEO of Hippocratic AI, a company building the first safety-focused large language model for healthcare.
A serial entrepreneur with a track record in AI and machine learning, he has founded and led multiple startups at the intersection of technology and healthcare, with previous ventures acquired by Google and Alibaba.
Marek Kowalkiewicz is Professor and Chair in Digital Economy at QUT Business School and a leading voice on the intersection of AI, business, and society.
Recognized by Thinkers360 as one of the Top 100 Global AI Thought Leaders, he is the author of The Economy of Algorithms: AI and the Rise of the Digital Minions, winner of the 2024 Australian Business Book Award in Technology.
Marek previously led innovation teams in Silicon Valley, established SAP’s Machine Learning Lab in Singapore, and held research appointments at Microsoft Research Asia.
His current work focuses on how algorithmic systems shape decision-making, redefine value creation, and challenge traditional notions of agency in business leadership.
Dr. Fatma Tarlaci is an engineering leader with a decade of expertise in AI. Formerly, she served as CTO and VP of Engineering at startups, where she led the development and deployment of robust AI solutions and led high-impact engineering teams.
As a technical advisor, she helps startups navigate AI adoption and productization, while also training the next generation of AI engineers and data scientists as an Adjunct Assistant Professor in Computer Science at UT Austin.
She recently stepped into the role of Chief AI Officer at SOAR AI, where she combines strategic leadership with hands-on development and helps guide the technical direction of their AI initiatives.
Munther A. Dahleh is the William A. Coolidge Professor of Electrical Engineering and Computer Science at MIT and the founding director of the MIT Institute for Data, Systems, and Society (IDSS).
His research explores decision-making under uncertainty, networked systems, and the economics of data, drawing on fields from control theory to distributed learning.
He leads cross-disciplinary efforts at MIT to understand how data and algorithms shape complex systems, from financial and power networks to social and neural systems.
Pascal Weinberger is an AI entrepreneur and investor whose work spans neuroscience-inspired machine learning, enterprise AI platforms, and moonshot innovation.
He began his career at the intersection of AI and neuroscience, later joining Google Brain and founding a successful AgTech AI startup.
He has led AI efforts at Telefonica’s Moonshot Factory, built enterprise-scale AI infrastructure at Augustus Intelligence, and now advises and invests in emerging AI ventures as a Venture Partner at AI Capital.
Saroop Bharwani is the founder of Senso and a longtime builder at the intersection of AI, human behavior, and enterprise systems.
With a background in computer engineering and neuropsychology, he has spent over a decade applying machine intelligence to predict and influence consumer behavior in large-scale environments.
His current work focuses on advancing the role of AI in shaping more adaptive and anticipatory financial systems.
Anirudh Narayan focuses on accelerating the adoption of agentic AI by bridging technical innovation with real-world application.
With a background in growth strategy and data-driven product development, he works at the intersection of AI deployment and business transformation, enabling organizations to unlock new forms of scale and autonomy through intelligent agents.
Dylan Hadfield-Menell is a leading researcher in AI alignment and directs the Algorithmic Alignment Group at MIT CSAIL. His work focuses on ensuring that agentic AI systems behave in ways that reflect human goals, values, and oversight, particularly in multi-agent and human-AI contexts.
As a Schmidt Sciences AI2050 Fellow, he is advancing methods to support the safe, beneficial, and trustworthy deployment of AI in the real world.
Tim Kraska is an Associate Professor at MIT CSAIL and founding co-director of the Data System and AI Lab (DSAIL). His research explores how machine learning can transform the foundations of data systems—improving performance, adaptability, and user interaction.
From rethinking core components with learned models to building intelligent interfaces for data science, his work enables more autonomous, accessible, and trustworthy AI systems at scale.