Observability in India has transformed dramatically over the past year. What began as a largely technical discipline focused on infrastructure resiliency and performance monitoring has evolved into a strategic business capability — and is now emerging as the essential control layer for the next wave of AI-driven autonomous operations.
“The observability landscape has completely changed in the last one year because of AI,” says Nalin Agrawal, Director of Solutions Engineering at Dynatrace. “Organisations have become far more mature in their adoption of AI, moving well beyond pilots.”
Historically, observability was the domain of CIOs and CTOs who used it primarily to ensure applications were running smoothly, reduce downtime, and maintain SLAs. That changed as business leaders began demanding deeper insights.
Agrawal recalls how one CIO from a large Indian bank pushed the conversation forward by drawing a powerful analogy: Just like in a taxi app, you can see exactly how much time the ride will take, I want my customers to track their personal loan or home loan application in real time — seeing it move from 10 days to 9 days to 8 days until completion.”
This demand for end-to-end business journey visibility led many large banks and insurance companies to connect disparate systems and create unified KPIs that reflect actual customer experience. The shift marked observability’s transition from a purely technical tool to a business decision-making platform.
The Autonomous Operations Era
The latest and most significant evolution is the rise of autonomous operations. As enterprises deploy increasing numbers of AI agents that interact with each other, consume data, and make independent decisions, observability is becoming the critical “control plane” that monitors, orchestrates, and governs these agents.“Autonomous operations is the new focus area,” Agrawal explains. “When AI agents are talking to multiple systems and taking decisions autonomously, someone needs to understand how these agents are behaving.
That layer today is observability.”Adoption varies significantly across sectors. Fintech companies and retail platforms (including payment gateways and food delivery services) have moved the fastest. A few have already achieved full “human-out-of-the-loop” automation, where AI systems analyse situations and take decisions independently. Traditional banks and insurance companies are more cautious, operating in a bi-modal fashion: 80% human-led with AI assistance in many areas, while running fully autonomous processes only for selected, lower-risk use cases.
Dynatrace has been evolving its platform to meet these new demands. A major milestone was the creation of Grail, its causal data lake, capable of ingesting and correlating up to one petabyte of data per day from across the entire technology stack — without traditional indexing.
The company has also adopted a hyper model AI strategy that combines three types of intelligence: Deterministic AI for consistent, explainable root cause analysis, Predictive AI for forecasting potential issues and Generative AI for natural language interaction
To bridge the gap between observability data and developer workflows, Dynatrace introduced the Model Context Protocol (MCP) server. This allows developers to query Dynatrace data directly from their preferred IDE or console using natural language prompts, eliminating the need to switch between multiple tools and hunt for logs.
Agrawal notes the efficiency gain: “Earlier, it used to take an average of four hours for a developer to get the relevant context and logs for an issue. MCP removes those multiple steps.”
Key challenges
Despite strong momentum, Indian CIOs continue to face several hurdles. Data fragmentation remains one of the biggest obstacles — large enterprises often have data scattered across hundreds of systems in different formats, making it difficult to prepare for effective AI use.
Other key challenges include building trust in AI recommendations, implementing strong governance and security guardrails (especially when agents can take autonomous actions), and managing the perception that one poor outcome could derail broader adoption.On the infrastructure front, many large organisations are still heavily investing in their own data centres rather than fully embracing public cloud, driven by security concerns and cost pressures. Interestingly, insurance companies have been more aggressive in moving to public cloud and SaaS, while some banks prefer to retain greater control through private infrastructure.
The Road Ahead
What is clear is that observability has moved from being a “nice-to-have” capability to a strategic must-have — especially as organisations pour significant investment into AI initiatives. Those that succeed will be the ones who can connect technical observability with business outcomes while building robust governance for autonomous systems.
As Nalin Agrawal summarises, the organisations that pull ahead will treat observability not merely as monitoring, but as the foundational layer that makes reliable, trustworthy autonomous operations possible at scale.