Splunk bets on agentic AI to deliver self-healing IT systems

Splunk is upgrading observability from a passive mirror to an intelligent, proactive partner

Modern IT teams have grown accustomed to dashboards full of metrics, logs and charts. They are the mirrors of digital operations, reflecting the health of applications and infrastructure. But what if those dashboards could do more than reflect? What if they could diagnose, decide, and even repair themselves? Splunk believes this is the next chapter in observability, and its new agentic AI upgrade for observability aims to bring self-healing IT systems closer to reality.

Splunk is embedding agentic AI into its Observability Cloud and AppDynamics, shifting observability from reflection to action.,The upgrade extends monitoring to AI agents and large language models, flagging drift, inefficiencies and runaway costs.,Positioned as the “intelligence layer” for enterprises, Splunk’s strategy highlights the convergence of AI, infrastructure, and business outcomes.

Why observability needs agentic AI

For years, observability platforms have been about clarity — giving engineers visibility into what is working, what is broken, and where the bottlenecks lie. In hybrid and multi-cloud environments, this visibility has been indispensable. Yet with the rise of AI agents, large language models (LLMs) and digital-first customer interactions, dashboards alone feel insufficient.

Splunk’s approach weaves agentic AI directly into its observability platforms. Rather than simply surfacing anomalies, the system analyses telemetry in real time, diagnoses root causes and recommends fixes. The ambition is not just to shorten incident response but to pre-empt problems altogether.

Kamal Hathi, SVP and GM of Splunk at Cisco, explains the leap:

“Agentic AI is reshaping what it takes for organisations to build and maintain a leading observability practice. We are delivering the only solution that can process, analyse and transform machine data from across all these environments into trusted inputs for LLMs, RAG pipelines, copilots and AI agents.”

“Agentic AI is reshaping what it takes for organisations to build and maintain a leading observability practice. We are delivering the only solution that can process, analyse and transform machine data from across all these environments into trusted inputs for LLMs, RAG pipelines, copilots and AI agents.”

Watching the watchers: observability for AI agents

The most intriguing dimension of Splunk’s upgrade is its extension into AI observability itself. Enterprises are no longer asking whether AI agents are running — they are asking whether those agents are performing well, securely, and cost-effectively.

With Splunk’s latest capabilities, IT leaders can measure whether an AI model is producing consistent outputs, staying within compute budgets, or showing signs of drift. If a model starts hallucinating or consuming GPU cycles beyond plan, Splunk detects and alerts in real time.

Patrick Lin, SVP and GM of observability at Splunk, frames the issue bluntly:

“As AI becomes more embedded in business operations, monitoring tools need to get smarter and provide real-time insights into whether models are delivering results efficiently and securely. Performance and cost have become critical metrics.”

“As AI becomes more embedded in business operations, monitoring tools need to get smarter and provide real-time insights into whether models are delivering results efficiently and securely. Performance and cost have become critical metrics.”

This matters in Asia, where financial services in Singapore, digital commerce in Indonesia, and government e-services in India are deploying AI agents at pace. A banking chatbot that drifts off script or a customer service bot that spikes compute costs is not a trivial nuisance; it directly affects margins and trust. For more on the regional impact, read about how AI is set to add nearly US$1 trillion to Southeast Asia's economy by 2030.

Infrastructure as the AI choke point

While AI agents get the headlines, the underlying infrastructure often determines success or failure. GPU shortages, cloud service quotas and accelerator costs are daily headaches for teams scaling AI workloads. Splunk’s move into proactive monitoring of infrastructure from resource bottlenecks to cost spikes; positions it as a guardian of the invisible plumbing.

This is particularly relevant for enterprises in Japan and South Korea, where demand for GPU clusters far outstrips supply. By flagging consumption issues early, Splunk hopes to help enterprises avoid outages and bill shocks. The World Economic Forum has highlighted the global challenge of AI infrastructure and resource scarcity.

The competitive lens

Splunk is not alone in its AI-powered observability mission. Datadog, Elastic Security, and Microsoft Sentinel are all investing in AI-enhanced detection and response. Datadog leans into behavioural analytics, Elastic into AI-driven search, and Microsoft into LLM-assisted security workflows.

Yet Splunk is carving out a distinctive position by embedding agentic AI triage; not just anomaly detection, but prioritisation and explanation of rare, high-risk alerts. This reduces analyst fatigue, a pain point echoed by security and IT teams across Asia-Pacific who face talent shortages and rising incident volumes. A similar trend is seen in the discussion around Will AI Agents Steal Your Job Or Help You Do It Better?.

Beyond logs: Splunk as the enterprise intelligence layer

Splunk’s ambition is broader than IT monitoring. Hathi suggests the company is positioning itself as the intelligence layer connecting infrastructure, AI, and business outcomes.

“Leaders often struggle with juggling a patchwork of tools that don’t always talk to each other, which can slow down teams and make it hard to get a clear picture of what’s going on. We are addressing this by creating a unified observability experience and using AI to accelerate problem detection and root cause analysis.”

“Leaders often struggle with juggling a patchwork of tools that don’t always talk to each other, which can slow down teams and make it hard to get a clear picture of what’s going on. We are addressing this by creating a unified observability experience and using AI to accelerate problem detection and root cause analysis.”

In other words, observability is shifting from an IT function to a cross-enterprise capability. As organisations in Asia scale AI adoption, the stakes are not merely uptime or latency, but customer satisfaction, regulatory compliance, and strategic agility. This transformation is part of AI's Secret Revolution: Trends You Can't Miss.

From reflection to resilience

Splunk’s message is clear: downtime is no longer acceptable in the AI era. In sectors where customer trust can evaporate in minutes — from fintech apps in Jakarta to logistics platforms in Shenzhen — a few moments of disruption can mean lost revenue and reputational damage.

Lin drives the point home:

“Observability isn’t just for ITOps and engineering teams. By sharing insights across teams, organisations can better align product development with real customer needs, improving satisfaction and driving business success beyond just technical performance.”

“Observability isn’t just for ITOps and engineering teams. By sharing insights across teams, organisations can better align product development with real customer needs, improving satisfaction and driving business success beyond just technical performance.”

Observability, then, is becoming less about dashboards and more about organisational resilience. Splunk is betting that agentic AI can keep enterprises a step ahead, ensuring IT systems not only run smoothly but actively repair themselves.

The big question is whether enterprises especially those in Asia’s fast-growing digital markets are ready to hand over the keys to agentic AI for observability. Would you trust an AI system to fix your IT before you even knew something was wrong?