The growing development, promise, and use of generative AI and agentic AI continue to drive dramatic change in the IT landscape of enterprises around the globe, from operations and infrastructure to software development, business management, and more.But perhaps one of the most impactful areas where agentic AI and AI are being felt is in the still-maturing field of IT operations and observability, where enterprises are gaining crucial insights into how their business applications are running across their IT environments and infrastructure.When integrated with traditional monitoring tools and observability platforms, agentic AI can enhance and revolutionize observability and IT operations by accelerating the aggregation and analysis of critical data from system logs, traces, and observability metrics. This enables deeper investigations into the root causes of system problems, faster and more thoroughly than existing tools and human engineers alone can. When integrated into observability systems, agentic AI enables engineers to scour massive amounts of data and dependencies across entire IT ecosystems, including microservices, servers, and databases. Engineers can use that data to conduct in-depth analysis of system properties, patterns, and problems using statistical and machine learning approaches, along with internal organizational data and use cases, to reveal new insights into system issues.How agentic AI is benefiting observability todayThaddeus Walsh, Principal Solutions Architect at Elastic, tells The New Stack that “Agentic data is an accelerator.”“It is an accelerator because I do not have to look at 20 pages of log lines or run some specific analysis process to familiarize myself with what the normal log signal looked like before an issue occurred,” Walsh says.“Agentic data is an accelerator… because I do not have to look at 20 pages of log lines or run some specific analysis process to familiarize myself with what the normal log signal looked like before an issue occurred.” — Thaddeus Walsh, Principal Solutions Architect at ElasticUsing agentic AI, time-consuming rote tasks can be offloaded to autonomous agents that complete them in a fraction of the time it takes human engineers.When AI agents are added to observability tools, IT teams gain powerful allies to support their business-critical work, says Walsh. The key to harnessing the power of AI is to first know your data, how it is retrieved, and how your processes and interfaces work, says Walsh. All these things are important to ensure that the tools you select are the correct fit for your company’s tasks and critical business workflows.How agentic AI is evolving for SRE teams and observabilityFor site reliability engineers (SREs), agentic workflows and AI observability tools can be used to accelerate RCA investigations, related code and system investigations, and quickly deliver approved, automated actions that provide useful insights and fixes.“I can ask an agent to look at the network data and tell me if there is any indication that an issue is associated with the network right now,” says Walsh. “The engineer can do their own work, then tell an agent to look at the feedback from other agents. The engineer can execute that work concurrently, which is more time efficient, and it allows the agents to provide consistency in how they inform the human operator.”This is a natural first step for bringing agents into the process to advise human operators on what to do next in the expanding investigations, says Walsh.Future evolutions of these tools will be even more powerful as agents are built directly into platforms to routinely, consistently, and accurately drive root cause and operational investigations with little human oversight, says Walsh. “We are not quite there yet, but I think that is where I think the industry will get to within the next 24 months.”Where agentic AI helps provide deeper observabilityThe addition of agentic AI capabilities is also driving broad innovations across observability tools and platforms for enterprise use cases such as: Debugging and improving application performance in DevOps. By using these tools, enterprises can unify and eliminate disruptive data silos and correlate metrics, logs, and distributed traces into a single observability pipeline to help engineers identify bottlenecks and streamline root cause analysis more quickly. Tightly integrating business and operational data to improve systems and productivity. This provides engineers with the critical ability to analyze all business data across operations to ensure its security, quality, and accuracy.Reducing the number of oversensitive operational alerts that are issued. This allows security engineers to focus on validated real threats rather than on benign “noise” that can provide false and deceptive clues, aimed at creating disruptive chaos to divert an engineer’s attention and enable more destructive attacks. Bolstering AIOps and AIOps observability. Using agentic AI tools democratizes access to broad and illuminating analytics and insights while significantly improving an engineer’s ability to identify and troubleshoot issues at scale.Using agentic AI-enabled observability tools, enterprises can also add more power and control to better manage their complex microservices architectures, observe real-time data across processes like order management, track and process business data across multiple silos, find anomalous system problems earlier, and improve the operations of their cloud-native and hybrid environments, including Kubernetes.Where to start using agentic AI in observabilityAn easy place to start is by creating an agent that sets up an alert and ticket or automatically sends a message to engineers when a service is down, says Walsh.“You can define how to make that decision and then offload it to an agent,” he says. “You can get a huge amount of noise reduction by implementing an agent that serves as sort of the quarterback of how communications are issued to people when an incident occurs.”Predictable, repetitive actions are also places where agentic AI can improve observability, he says. “Agentic AI brings to the table an ability to bring a human likeness into a process.”“You can get a huge amount of noise reduction by implementing an agent that serves as sort of the quarterback of how communications are issued to people when an incident occurs.”Another example is when many alerts come in reporting hundreds of impacts from an incident that must be investigated and evaluated for problems, says Walsh. “I do not want to receive 400 emails with discrete little signals. I want to see a summary of those signals. And I want to be told what signals I have already checked.”The value of agentic AI in these processes lies in its being reactive and always on, says Walsh. “The alerts start appearing, and immediately the agent goes into action.”What’s next: Democratizing observability data By combining the speed and efficiency of observability with the power of agentic AI, enterprises gain deeper insights, more accurate real-time responses, and more. And while observability data has long been the domain of SREs and IT operations teams locked into a single area of the business, that pattern is changing.Now agentic AI and generative AI are making it possible for developers, DevOps engineers, application owners, security analysts, and business stakeholders to tap into observability data directly, ask questions in plain language, and get real answers without needing to be experts. See also: Most enterprises will hand root cause analysis to AI agents within two years“Those are natural steps along the yellow brick road” of tomorrow’s observability capabilities, says Walsh. Learn more about democratizing observability in Elastic’s recent webinar with Walsh, Senior Principal Solutions Architect Brad Quarry, and TNS host Alex Wilhelm.The post Agentic AI in observability: accelerating root cause analysis appeared first on The New Stack.