Chloe Maraina is a distinguished Business Intelligence expert with a profound knack for translating complex big data into actionable visual narratives. With a background rooted in data science and a forward-looking perspective on system integration, she provides a unique vantage point on how the next generation of observability is being reshaped by agentic AI. In this discussion, we explore the evolving landscape of automated software delivery, the transition from manual triaging to intelligent root cause analysis, and the strategic importance of industry partnerships in a world where data is increasingly federated and dynamic.
The conversation covers the shift away from hardcoded API integrations toward fluid, agentic connections, the operational impact of new SRE-focused AI tools, and the practicalities of using no-code platforms to customize telemetry analysis.
Many organizations choose to partner for CI/CD and software delivery automation rather than building a proprietary, all-in-one stack. How does this partnership-heavy strategy impact the integration speed between observability data and platforms like GitHub? Please explain the step-by-step logic of how agents now handle these connections without hardcoded APIs.
The beauty of a partnership-heavy strategy is that it allows experts to stay within their domain while creating a more flexible ecosystem for the end user. In the past, connecting observability data to a CI/CD pipeline required engineers to spend weeks or months building and maintaining rigid, hardcoded API connection points for every single tool in the stack. Now, we are seeing a shift where agents are “smart” enough to understand the context of an infrastructure change or a configuration update without a manual roadmap. When an issue is identified, the observability agent identifies the specific change and dynamically pushes that information to a partner agent on the GitHub or Amazon side. It’s a fluid, conversation-like interaction between systems where the agents negotiate the data exchange themselves, eliminating the traditional bottleneck of custom coding.
Modern incident management often involves specialized agents analyzing observability data alongside tools like ServiceNow or PagerDuty. How does this shift from manual triaging to automated root cause analysis change the daily workflow for a site reliability engineer? Share specific metrics or anecdotes regarding the impact on resolution times.
For a Site Reliability Engineer (SRE), the traditional workflow was often a frantic race against the clock to correlate disparate logs and dashboards while stakeholders demanded updates. With the introduction of specialized SRE agents, the heavy lifting of “detective work” is handled by the AI, which can simultaneously query platforms like ServiceNow or Atlassian to ask what services are impacted and which users are suffering. Instead of manually digging through telemetry, the SRE receives a summarized investigation that points directly to the source of the failure. This creates an environment where the agent acts as an automated triage partner, surfacing recommendations that can turn hours of manual investigation into minutes of verified response. By focusing on automated root cause analysis, teams can stop the repetitive cycle of “war rooms” and move toward a more proactive posture where the system essentially explains itself.
No-code AI agent builders allow users to refine analysis for their specific internal systems rather than relying on general-purpose platforms. What are the practical steps for a team to customize an agent for their unique telemetry data? Explain how this customization helps surface recommendations for automating repetitive workflows.
The practical path to customization begins with feeding the no-code builder specific parameters of your internal environment, rather than relying on the broad, generic logic of a standard AI model. A team starts by defining the specific telemetry markers—whether those are custom metrics or unique log patterns—that signify healthy operations for their unique services. The agent then uses this refined context to analyze data through a more narrow, relevant lens, which allows it to spot anomalies that a general-purpose tool might ignore. Because the agent understands the specific nuances of the user’s system, it can suggest precise automation for repetitive workflows, such as auto-scaling resources when a specific bottleneck is predicted. This level of customization ensures that the AI’s recommendations are not just theoretical, but are grounded in the actual operational reality of that specific business.
Observability platforms are expanding to include federated logs on Amazon S3 and eBPF network metrics. How do these technical features translate into better business impact analysis for AI applications? Describe the process of correlating infrastructure health with actual user experience and financial outcomes.
Expanding observability to include federated logs on S3 and eBPF metrics provides a granular look at the “connective tissue” of modern applications, which is vital for understanding business impact. By pulling data from these diverse sources, a platform can create a clear line of sight from a packet drop at the network level all the way to a failed transaction in a shopping cart. For AI applications, this means you can track the cost and performance of specific LLM calls against the actual user satisfaction scores in real-time. We are moving away from static dashboards and toward dynamic homepages that correlate infrastructure health directly with financial outcomes, showing executives exactly how a server delay impacted the bottom line. This technical depth allows the system to prove its worth by translating technical glitches into the language of lost revenue or diminished user retention.
Executives often compare new agentic ecosystems to the traditional root cause analysis tools they have used for a decade. How can organizations move past the “wait and see” phase to prove the value of agentic AI in a live production environment? Provide examples of what success looks like.
To move past the skepticism, organizations need to stop viewing agentic AI as a science project and start deploying it against specific, high-friction pain points in their live production environments. Success doesn’t look like a total replacement of the human engineer; rather, it looks like a “co-pilot” scenario where the AI successfully identifies a cascading failure before the first customer ticket is even submitted. For instance, an executive knows the system is working when an agent detects a memory leak in a new microservice and automatically alerts the specific developer responsible with a pre-packaged analysis of the code error. Proving value requires showing that these agents provide exponentially more insight than the tools of the last 10 years by being proactive rather than just descriptive. When an organization sees a measurable drop in Mean Time to Resolution (MTTR) and a reduction in “alert fatigue” for their staff, the transition from “wait and see” to full adoption happens very quickly.
What is your forecast for agentic AI observability?
I forecast that we are entering an era of “open agentic collaboration,” where the true winners will not be the companies that try to build a closed, all-in-one ecosystem, but those that facilitate seamless communication between agents across different vendors. We will see a shift from the single-pane-of-glass dashboard to a “notebook” style of interaction, where AI agents and humans collaborate in a dynamic document to solve complex architectural problems in real-time. Over the next few years, the focus will move entirely away from just collecting data to orchestrating intelligent actions, making the observability platform the central brain of the entire enterprise. As these systems prove themselves in high-pressure environments, the “agentic makeover” will become the standard requirement for any organization that wants to remain competitive in a cloud-native world.
