Partnering with Empirik: Building the Autonomous Infrastructure Engineer

This one started close to home. Before Sequoia, our Chief Digital and Information Officer, Avon Puri, had spent two decades running IT and infrastructure at scale as SVP & CIO at Rubrik, and VP of Engineering and Applications at VMware (also my alma mater in the tech world). Avon and Sudheer Dhurjati, who worked alongside him at all three companies, watched the same story repeat: enterprise infrastructure operates in a reactive posture. Companies spend tens of billions of dollars on observability—metrics, logs, and traces—to tell us when a system is failing. But traditional observability is inherently post-mortem.  

The real driver of infrastructure behavior is change: a new identity policy is attached, a new Terraform provider completes, or a feature flag flips in production. However, change has been treated as an administrative process constraint, a static ticket, an approval form, or as tribal knowledge buried in email or Slack. It was sometimes recorded for compliance, but never truly put to work to do actual infrastructure engineering jobs. 

If you understand the real-time stream of changes across every layer of the environment, you don't just get better alerts—you unlock true autonomy. Every change reflects an engineer's intent; once a system can interpret that intent and assess its impact, it can decide whether to act on its own or loop in a human.  By bringing agentic capabilities to infrastructure, you solve the full lifecycle of infrastructure engineering jobs, shifting from a reactive firefighting posture to a proactive one.

That insight became the founding pillar behind Empirik. Avon and Sudheer built the initial version inside Sequoia. We saw companies rushing to deploy AI agents to automate coding, marketing, and many other enterprise processes, but infrastructure was the one place nobody was modernizing. After the same rigorous evaluation any other Sequoia investment goes through, Empirik spun out as an independent company. And now, it’s under the leadership of CEO Kartik Chandrayana, a repeat founder who sold his first company, Twin Prime, to Salesforce, then spent five years as Salesforce's VP of Product for Observability and Big Data, and most recently was CPO at Quantum Metric. Today, we're excited to share that Empirik is emerging from stealth with $21M in funding.

Why change, and why now

Agentic coding has taken off because code lives in version-controlled repositories that make a system's architecture and intent legible to a model. Infrastructure has never had that foundation. Infrastructure-as-code is a good start, but because it represents a declared state and not what's actually running in production, it's inherently behind. There's no unified, real-time representation of an organization's infrastructure across the maze of cloud resources, Kubernetes, networking, identity, and legacy systems. Infrastructure does not have the notion of a pull request; instead, every change still runs through tribal knowledge and institutional memory of people rather than the machine rigor of software. 

Observability is already a large, proven market—Datadog, Dynatrace, and Splunk are behemoths with billions in revenue. But all of observability is built to measure symptoms: latency spiked, errors climbed, a pod crashed. Empirik's bet is that there's a fourth pillar hiding in plain sight—the changes themselves. Instead of "latency in service X increased 300% in the last 24 hours," Empirik tells you, "Joe deployed a Terraform provider that shut down three VMs in AWS"—the actual cause and context, not just the symptom. And why stop there? With Empirik, you can actually warn or stop Joe even before he deploys Terraform because Empirik computes that risk at the inception of change.

Building the infrastructure compiler

As Kartik describes it: infrastructure today is where software engineering was three years ago—before Claude Code and Cursor, when every action needed a human in the loop because there was no reliable map of the system to reason over. Empirik is building that map, and putting it to work as the Autonomous Infra Engineer—AI that can safely plan, reason about, and operate enterprise infrastructure. It's already plugged into agentic workflows via MCP, connecting directly into tools like Claude and ChatGPT, so that humans and agents alike can reason over infrastructure change with real context—not guesswork.

Under the hood is an infrastructure compiler. Just as a traditional compiler translates source code into executable software, Empirik converts metadata from a customer’s clouds, identity providers, CI/CD, and SaaS systems into a live, unified operational graph—mapping thousands of relationships between assets and accounts. That graph is how the product knows, in real time, what changed, who changed it, and whether it’s the reason something broke.

What I particularly appreciate about the team’s approach is that they didn’t take shortcuts. It’s easy for modern startups to build exclusively for greenfield Kubernetes or a single public cloud provider. But from day one, Avon, Sudheer, and Kartik designed Empirik for the reality of the Fortune 500 enterprise: a complex, multi-decade hybrid stack spanning public clouds, legacy on-prem infrastructure, identity layers, and critical SaaS apps. An infrastructure graph that only sees half your environment misses the mark for an enterprise buyer. Empirik’s graph connects the entire surface area so teams and agents have complete operational context.

The critical question, answered

The traction backs up the thesis. Fortune500 Financial enterprise’s interest in the product traces directly back to a P0 outage caused by an unapproved change. TCBPay, a payment processing company with $1.5B+ in annual volume now catches sensitive config file changes in 10 seconds, down from 30 minutes. Avahi, an MSP, integrated Empirik within their GitHub Actions to evaluate and block risky changes at the source. Guardant Health found Empirik so valuable that it expanded its deployment from production to all dev environments, and even expanded to on-prem. And a global Fortune-50 CPG company is also expanding its footprint, relying on Empirik to drive most of its agentic and autonomous infra roadmap. Empirik is now processing millions of raw change and telemetry events every week, over dependency graphs spanning a few million resources across all major clouds and on-prem systems. Customers are reporting benefits in multiple domains: preventing risky changes from making it to production, detecting unapproved changes, faster incident response, making rollback decisions, and detecting drifts, among others.

The success is no surprise given this team. Avon and Sudheer are two of the best operators we know, and they built Empirik the way they'd have wanted it built for themselves. Kartik, meanwhile, has firsthand expertise in shipping observability products at enterprise scale. Most enterprises still can't answer that simple but critical question: what changed, and did that cause the incident? Empirik is building the system of record for that question—for humans today, and for the agents that will increasingly be making those changes themselves tomorrow.