Tanium’s addition of the new
Ask Agent to its autonomous endpoint management (AEM) platform is the latest example of the cybersecurity industry embracing agentic AI.
This month, the California-based vendor unveiled Ask Agent to autonomously pull in vast amounts of telemetry, put it into context, create insights, and make real-time decisions to protect and secure endpoints.
In a
statement,
Tanium co-founder and Executive Chairman
Orion Hindawi said that “IT needs agentic AI – agents that understand each specific IT environment, provide thoughtful change management and that automate comprehensive and intelligent QA with automatic rollback if failures occur.”
Tanium Not Alone in Its Thinking
Lucia Stanham, product marketing director for
CrowdStrike, said agentic AI is a significant advancement from traditional AI, which relies on predefined rules or passive responses and responds to specific inputs, and generative AI, which goes beyond those predictive capabilities to create content based on the data it’s trained on.
“Agentic AI is a big leap forward in the development of AI, adding the capability to operate with a level of autonomy,” Stanham
wrote in a blog post. “It can assess situations, adjust strategies, and pursue objectives in dynamic environments, all while refining its own approach over time. Agentic AI is once again significantly changing the way that humans interact with AI.”
Tanium, like many similar vendors, has been aggressive in its adoption of AI as a tool to improve endpoint management and security. In March, the company released a
report that showed automation significantly improves security and reduces human errors. In a survey of Australian IT teams, 78% said that automation tools can improve security by shortening patch cycles, reducing vulnerability exposure, and improving incident response.
AI Agent Adoption is Accelerating
The Ask Agent integrates AI-driven workflows into Tanium’s AEM platform, accelerating tasks like software management, data discovery, and the summarization of dashboard data to give admins a single place to investigate, troubleshoot, and remediate issues.
Cybersecurity vendors have for years been incorporating AI into their operations and tools, with the global market hitting $22.4 billion in 2023 and being expected to
reach $60.6 billion by 2028, growing an average of 22% a year, according to MarketsandMarkets.
The use of agentic AI for cybersecurity is still a relatively new area, but market research firms are beginning to put some numbers to it as well. Analysts with Market.US expect the space to jump from $738.2 million in 2023 to almost
$21 billion by 2034.
“Integrating agentic AI into cybersecurity has catalyzed a burgeoning market, with organizations worldwide investing in advanced AI-driven security solutions,” the Market.US researchers wrote. “This market encompasses a range of products and services designed to autonomously detect and respond to cyber threats, thereby reducing reliance on human intervention.”
Humans, Guardrails Needed
Rob Enderle, principal analyst for
The Enderle Group, told MSSP Alert that having tactical AI like Ask Agent is important to respond at machine speeds to the ever-increasing number of AI-driven malware and cyberattacks. That said, it’s crucial that the necessary guardrails are in place and that humans continue to be included in the process.
“Strategy and the AI’s guard rails should still be set by humans, and there may need to be a second AI that assures the operation of the primary security AI, as should it become compromised, it could do massive damage,” Enderle said. “If there isn’t sufficient direction or protection over this endpoint AI, it could become compromised and do significant damage, perhaps even more damage than a typical attack, given it would have massive privileges to do its job effectively.”
MSSPs Are in the Agentic AI Mix
Given that, it not only needs human oversight, but also a second AI tool that assures it doesn’t become compromised. That goes for MSSPs that will use Tanium’s agent or other agentic AI products.
“The only additional thing [with MSSPs] is to ensure any corruption of the platform doesn’t spread between clients,” the analyst said. “AIs tend to exchange training data, but should one company’s data become compromised. It could compromise all of the MSSPs’ and MPS’s clients.”
Tanium said its Ask Agent technology included human-in-the-loop oversight for control and accountability.
MSSPs and MSPs, like security vendors, are deploying agentic AI to bolster their services. Recently, CrowdStrike and its Red Canary subsidiary in August
announced a partnership with
Zscaler to bring AI agents and other tools that security operations centers (SOCs) can use. Earlier that month, DXC Technology, in partnership with 7AI, unveiled a
new SOC model built on AI agents.
F5 in June
bought startup Fletch to bring AI agents into its Application Delivery and Security Platform.
It’s good that security firms are embracing agentic AI, but Enderle warned that “there is a severe shortage of people who understand and can use this technology safely.”
