For decades, modern enterprise IT has operated under an inescapable paradox: while software engineering has continually evolved through rigorous version control, automated testing, and streamlined deployment pipelines, enterprise infrastructure has remained largely reactive, trapped in a post-mortem cycle of firefighting. Companies routinely invest tens of billions of dollars into advanced observability tools—relying on complex metrics, logs, and traces—merely to tell them when a critical system is failing.
That fundamental imbalance is now facing a systemic challenge. Empirik, an infrastructure intelligence startup incubated within Sequoia Capital, is officially emerging from stealth today, announcing $21 million in funding. The company aims to solve the full lifecycle of enterprise infrastructure engineering by introducing agentic AI capabilities designed to shift organizations from a reactive posture to a proactive one.
Led by Chief Executive Officer Kartik Chandrayana alongside co-founders Avon Puri and Sudheer Dhurjati, Empirik seeks to address a blind spot that has persisted across the technology sector even as enterprises race to deploy artificial intelligence agents to automate everything from coding workflows to marketing operations. While software development has embraced automation because code repositories make architecture and intent legible to machine learning models, infrastructure has historically lacked a unified foundational layer.
The genesis of Empirik traces back to deep industry experience. Before launching the company, Chief Digital and Information Officer Avon Puri spent two decades running enterprise IT and infrastructure at scale, serving as SVP and CIO at Rubrik and VP of Engineering and Applications at VMware. Alongside Sudheer Dhurjati, who worked directly with him across multiple major technology companies, Puri repeatedly witnessed the same operational friction: enterprise infrastructure operating strictly in response to failure rather than managing the root causes of change.
Why Change, and Why Now
According to the founders, the true underlying driver of infrastructure behavior is not failure itself, but change. Every incident in a production environment is preceded by a modification: a new identity policy is attached, a new Terraform provider completes its execution, or a feature flag is flipped. Historically, however, change has been treated merely as an administrative constraint—a static ticket, a manual approval form, or tribal knowledge buried deep within Slack threads and email chains. While these records are occasionally archived for compliance audits, they have rarely been utilized to perform actual infrastructure engineering work.
The core premise behind Empirik is that understanding the real-time stream of changes across every layer of an enterprise environment unlocks true system autonomy. Every infrastructure change is a physical reflection of an engineer’s intent. Once an autonomous system is equipped to interpret that intent and accurately assess its impact, it can autonomously decide whether to execute the change safely or loop in human oversight.
This philosophy directly challenges traditional observability markets. While established software behemoths like Datadog, Dynatrace, and Splunk have built multi-billion-dollar empires by measuring operational symptoms—such as a sudden latency spike, an accumulation of errors, or a crashed server pod—Empirik focuses on uncovering what it calls the fourth pillar of infrastructure intelligence: the changes themselves.
Instead of generating a standard alert noting that service latency increased by three hundred percent over a twenty-four-hour period, Empirik aims to surface the root context directly, identifying precisely which configuration or deployment triggered the event. Furthermore, by computing operational risk at the very inception of a change, the platform can warn or automatically block engineers before a risky deployment ever reaches production.
Building the Infrastructure Compiler
Describing the current state of the industry, CEO Kartik Chandrayana notes that enterprise infrastructure today mirrors where software engineering stood three years ago—prior to the advent of advanced coding assistants, when every administrative action required human intervention because no reliable systemic map existed to reason over. Empirik is constructing that foundational map, deploying it as an autonomous infrastructure engineer capable of planning, reasoning about, and safely operating complex enterprise environments.
The platform is natively integrated into agentic workflows via the Model Context Protocol, connecting directly into modern developer tools like Claude and ChatGPT so that both human operators and autonomous agents can reason over infrastructure changes utilizing real-time operational context rather than guesswork.
Underneath the platform lies what the company describes as an infrastructure compiler. Much like a traditional software compiler translates human-readable source code into executable software, Empirik’s compiler ingests and converts metadata from a customer’s multi-cloud environments, identity providers, continuous integration pipelines, and software-as-a-service applications into a live, unified operational graph. This comprehensive graph maps thousands of dynamic relationships across enterprise assets and user accounts, allowing the system to instantly identify what changed, who initiated the change, and whether that modification contributed to an active system failure.
Crucially, the platform was designed from inception to address the operational realities of Fortune 500 enterprises. Rather than focusing exclusively on greenfield Kubernetes deployments or a single public cloud provider, the founding team built Empirik to support complex, multi-decade hybrid technology stacks spanning public clouds, legacy on-premise infrastructure, identity management layers, and critical corporate applications. By connecting the entire operational surface area, the platform ensures that both engineering teams and AI agents maintain complete context.
Proven Enterprise Traction
The market thesis underpinning Empirik is already reflected in early customer adoption across diverse industries. Interest from Fortune 500 financial institutions often stems directly from previous operational disruptions caused by unapproved configuration changes. For instance, payment processing firm TCBPay, which handles over $1.5 billion in annual transaction volume, now detects sensitive configuration file modifications in ten seconds, down from a previous manual window of thirty minutes.
Managed service provider Avahi integrated Empirik directly into its GitHub Actions workflows to evaluate and automatically block risky changes at the source. Meanwhile, Guardant Health expanded its initial deployment from production environments to cover all internal development workflows and on-premise systems, while a global Fortune 50 consumer packaged goods enterprise is utilizing the platform to drive its autonomous infrastructure roadmap.
Processing millions of raw change and telemetry events every week across dependency graphs that span millions of individual resources, Empirik’s early users report significant operational improvements. These include preventing risky deployments from reaching production, detecting unmanaged configuration drifts, accelerating incident response times, and guiding automated rollback decisions.
By addressing the fundamental question that continues to challenge enterprise IT organizations—identifying precisely what changed and whether that change caused an incident—Empirik is establishing a new system of record designed to support human operators today and the autonomous agents that will manage those systems tomorrow.
Comments (0)