Operant AI Launches AI Gatekeeper for Autonomous AI Deployments

Operant AI, headquartered in Silicon Valley, has introduced AI Gatekeeper, a real-time security application designed for live AI deployments, including autonomous agents and agentic AI workflows. The solution supports operations across Kubernetes, private cloud, hybrid, and edge environments.

As organisations increasingly adopt autonomous AI agents and multi-agent workflows—particularly in high-growth regions such as India—security challenges have become more prominent. According to Deloitte’s State of GenAI report, over 80% of Indian organisations are exploring the use of autonomous agents, with 50% implementing multi-agent setups that require minimal human involvement.

AI Gatekeeper Capabilities

AI Gatekeeper builds on Operant AI’s existing 3D Defence framework and introduces new capabilities, including trust scoring for AI agents, agentic access control mechanisms and threat detection and blocking for Model Context Protocols (MCPs) and Non-Human Identities (NHIs). These features aim to mitigate the risks posed by rogue agents in live AI systems.

Third-Party Vendor Dependence Elevates Risk

Recent engagements between Operant AI and Indian enterprises highlight increasing interest in agent-based AI deployments. However, there is a growing reliance on third-party vendors, which introduces additional security concerns such as data leakage, model poisoning, and unregulated agent behaviour.

AI Gatekeeper is designed to address these risks by securing agentic AI deployments at runtime. It provides a governance framework for organisations to manage data security, monitor AI behaviour, and maintain control across various deployment platforms.

“The AI that we are now securing is a completely new beast compared to even two years ago,” said Vrajesh Bhavsar, Operant AI’s CEO and co-founder. He added that today, RAG applications to AI agents and AI inference systems operate at a completely new scale, which is why AI can’t be secured in isolation. AI Gatekeeper can bring Operant’s unique defensive capabilities to everywhere customers are deploying AI, alongside critical new capabilities for protecting sensitive data and the rest of the application environment from the new attack surface that is being fueled by rapid Agentic AI adoption.”

“We are seeing three trends happening right now: First, incredibly fast deployment of AI models and AI Agents for novel use cases; second, adoption of new platforms beyond the traditional cloud providers; and lastly, the requirements and responsibilities for security, infrastructure, data infosec and AI converging,” said Raj Yavatkar, CTO of Juniper Networks.

Operant AI Enhances Security for Evolving AI Workflows and Platforms

Operant AI has developed a security solution designed to protect critical business transformations while enabling AI-native teams to scale innovation securely. The newly launched AI Gatekeeper supports this objective by addressing emerging risks associated with autonomous AI deployments.

AI applications and agents are increasingly being deployed across both traditional and non-traditional platforms. In addition to cloud hyperscalers such as Amazon EKS, Fargate, Bedrock, Microsoft Azure, and Google Cloud, enterprises are expanding AI workloads to platforms including Databricks, Snowflake, and Salesforce.

This shift brings the AI ecosystem closer to the core data sources, increasing the complexity and scope of security requirements.

Rising Threat Exposure in Agentic Workflows

As enterprises integrate agentic AI workflows, the risk of security incidents grows. These workflows require built-in controls and safeguards to mitigate emerging failure modes associated with autonomous systems. Frameworks such as the Model Context Protocol (MCP) introduce new vectors of exposure, including vulnerabilities like tool poisoning, which demand a different security approach than conventional models.

The evolving nature of AI deployments calls for real-time, context-aware security tools. AI Gatekeeper addresses this need by securing agentic workflows across diverse environments and platforms, aligning with the growing demand for AI governance, threat detection, and risk mitigation.

“Securing AI Agents is a critical priority for AI-native companies because you can’t hand off that level of autonomy at scale to these systems without appropriate controls in place,” said Martin Choluj, CISO of Clickhouse.

Operant’s AI Gatekeeper launch follows its inclusion as a representative vendor in Gartner’s Market Guide for AI Trust, Risk, and Security Management (AI TRiSM) and mention in Gartner’s recent research note, “How to Secure Custom-Built AI Agents.”

Operant Expands AI Gatekeeper Capabilities to Strengthen AI Runtime Security

Support for Multi-Cloud Environments and AI Platforms

Operant has introduced new capabilities in its AI Gatekeeper offering, extending runtime protection for AI workloads across public, private, and hybrid cloud environments. Building on its 3D Runtime Protection system, the updated platform now offers:

• Compatibility with environments beyond Kubernetes.

• Live catalogs that automatically track and update AI workloads, agents, tools, and models in use across an enterprise. Supported platforms include OpenAI, Deepseek, Cohere, Anthropic, Hugging Face, and others.

• Expanded support for large data platforms, large language model (LLM) platforms, and AI agent ecosystems.

• Analytics that provide visibility into all deployed defenses and threats mitigated in real time.

Cross-Platform Threat Modeling for AI Workloads

AI Gatekeeper now includes comprehensive threat modeling tools to support security across diverse environments. Key features include:

• AI Security Graphs that map and flag high-risk data flows between AI workloads, agents, and APIs across multiple platforms.

• Built-in mappings to OWASP Top 10 threat vectors for AI/LLMs and AI agents, such as sensitive data leakage, credential exposure, prompt injection, and model/data poisoning.

• Detailed insights into active threats and the specific workloads or APIs affected.

Advanced Threat Detection Capabilities for AI Agents

To secure AI agents and workflows, Operant has introduced advanced detection and control mechanisms:

• Identification and mapping of supply chain risks using trust scores and security boundaries.

• Detection and blocking of unauthenticated or unauthorized AI agents.

• Implementation of least privilege runtime execution and minimal-permission boundaries for AI agents to reduce risk.

Support for MCP and AI Non-Human Identities (NHIs)

Operant has added protections specific to Model Context Protocol (MCP) and AI Non-Human Identities (NHIs), including:

• Defense mechanisms at both the runtime and API access layers for agent tools built using MCP.

• Broader identity and access control features for AI NHIs to ensure secure and controlled interactions.

Read More:

Integrating AI, data analytics, and Technologies Help Businesses

We Emphasise on Interdependence of AI and Cloud technologies

AI, Security, and Quantum Computing Beholds the Future

Focus on Hybrid Cloud Integration and AI powered IT Automation