Coralogix is expanding its observability platform by adding tools that will help organizations and MSSPs manage what the company’s top executive calls the “gray area” of AI that can hinder the effective use of the emerging technology.
The Boston-based vendor this week rolled out its AI Center, which includes customizable evaluators that can detect when an organization’s AI solutions seem to be working correctly but actually are having issues related to responses, such as whether an exchange contains toxic content, the AI is hallucinating, or a hacker is trying to access the chatbot to steal data.
“AI introduces risks traditional observability tools can’t track,” Coralogix co-founder and CEO Ariel Assaraf told MSSP Alert. “Unlike standard observability, which follows clear right-wrong signals, AI operates in a shifting gray area.”
The AI Evaluation Engine in the AI Center closes the gap by continuously monitoring models, assessing performance, and identifying prompt and response issues, Assaraf said, adding that “this engine goes beyond infrastructure monitoring. It has AI evaluators that, when activated, detect and flag hallucinations, data leaks, and all security vulnerabilities in real-time.”
AI Adoption and Security
Enterprise adoption continues to accelerate, with global consultancy McKinsey and Co. noting that last year, 78% of survey respondents said their organizations use AI in at least one business function, up from 72% in early 2024 and 55% percent in 2023.
Given the expanding adoption of AI, security becomes critical, according to Exabeam.
“As AI technologies become increasingly integrated into various aspects of digital infrastructure, the importance of cyber security measures tailored to these technologies grows,” the Foster City, California-based cybersecurity firm wrote. “The goal is to prevent unauthorized access, manipulation, or misuse of AI systems, which could lead to privacy breaches, misinformation, or other forms of cyberattacks.”
Not Just Another Piece of Software
Coralogix’s full-stack observability platform is designed to monitor and manage data, application performance, security, and infrastructure in real time and includes such tools as SIEM, real user monitoring, and governance. Most companies treat AI as simply another piece of software that they can address with traditional security tools, Assaraf said.
“This approach ignores the reality that AI requires an entirely new infrastructure to handle its evolving risks,” he said. “Standard security tools might catch data leaks and adversarial attacks, but they fail to address the unique issues that AI agents pose, from hallucinations and incorrectness to toxicity to compliance issues.”
Keeping Tabs on AI
As an example, the CEO noted that AI systems are vulnerable to a new form of denial-of-service attack that targets token consumption that drains an organization's resources by forcing AI models to process excessive input. DoS attacks are well-monitored in traditional applications, but such AI-specific resource-exhaustion attacks are often ignored.
AI Center will catch such DOS attacks and other threats for its more than 2,000 enterprise customers, he said.
Along with the AI Evaluation Engine, AI Center comes with AI security posture management that includes dashboards monitoring the security and performance of AI agents. There also are tools for tracking user interactions to detect issues like suspicious resource consumption and cost harvesting attempts while optimizing budgets, as well as performance metrics to find problems like poor response accuracy, latency spikes, and malicious user inputs.
A Much-Needed Tool for MSSPs
Such capabilities around real-time threat detection, rapid response, and cost-efficient security analysts are important for MSSPs, Assaraf said, adding that Boston-based Coralogix helps MSSPs scale by reducing storage costs by 70% without compromising on the amount of data they can observe while increasing the speed and accuracy of monitoring and compliance.
Coralogix in 2022 launched its own managed service, Snowbit, that includes experts monitoring customers in real time and combines SIEM, cloud security posture management, and managed detection and response on a single platform.
“Additionally, we enable MSSPs such as Optiv to leverage our platform to enhance their own security services, providing better protection for their customers,” he said. “MSSPs and MSPs can now offer AI-specific observability and security services that weren’t possible before. Up until now, CISOs and security managers were blind to the AI being utilized within their environment.”
Coralogix’s AI Center is the result of its acquisition in December of Aporia, a startup that offered AI observability capabilities and AI guardrails, according to Coralogix co-founder and CTO Yoni Farin. At the time of the deal, Coralogix executives said they planned to launch a dedicated AI research center to focus on such fundamental problems in AI as transparency, security, governance, and control, with expectations of investing tens of millions of dollars in the initiative over the next two years.
