Sequoia Capital is spinning out Empirik, an artificial intelligence startup seeking to change how companies manage technology infrastructure by using AI agents to detect and prevent system failures before they trigger costly outages.
The company announced Tuesday that it has raised $21 million in seed funding from Sequoia, Canapi and Alumni Ventures as it launches as an independent business.
Empirik was conceived by Sequoia chief digital and information officer Avon Puri and Sudheer Dhurjati, another senior technology executive at the venture capital firm. Both had extensive experience managing large-scale infrastructure, including Puri’s more than decade-long work overseeing infrastructure at Rubrik and VMware.
Register for the next Tekedia Mini-MBA.
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
The founders began exploring the idea about three years ago, as advances in large language models suggested that AI could do more than assist engineers with troubleshooting. They believed AI could analyze changes across complex technology environments and identify potential failures before they occurred.
That approach became the foundation for Empirik, which monitors changes to an organization’s infrastructure and attempts to determine how a change in one part of a system could affect other components.
The startup was incubated by Sequoia in 2023 before the firm recruited Kartik Chandrayana as chief executive earlier this year. Chandrayana previously served as chief product officer at Quantum Metric and held an observability leadership role at Salesforce.
Empirik is entering a market that has become very relevant as companies add more software, cloud infrastructure, and AI-generated code to their technology stacks. The faster development cycle created by AI coding tools can also increase the number of infrastructure changes engineers must review and manage. That creates a potential bottleneck for DevOps and site reliability engineering teams, which are responsible for keeping applications and underlying systems operational.
“There has always been a lot of money spent in keeping systems up and running,” Sequoia partner Bogomil Balkansky told TechCrunch.
He said many existing observability products struggle to understand the dependencies linking complex systems.
Empirik is designed to operate as an autonomous layer over infrastructure, analyzing changes and assigning different levels of risk. Low-risk changes can be allowed to proceed, larger changes can trigger additional safeguards, while potentially dangerous updates can be escalated to engineers for review.
Balkansky described the system as an autonomous “traffic cop” for infrastructure.
That description marks an exception as companies increasingly move toward AI-assisted software development. Coding agents can dramatically increase the amount of software engineers are able to produce, but every additional application, code change, or deployment can create new dependencies and potential points of failure.
Empirik is therefore betting that the next bottleneck in AI-driven software development will not necessarily be writing code, but ensuring that the infrastructure supporting that code remains stable. The startup has already attracted customers ranging from early-stage companies to several Fortune 500 businesses, including S&P Global, Guardant Health and a major consumer packaged goods company.
Chandrayana said the company’s ambition is to bring the same type of productivity gains to infrastructure engineering that AI coding tools such as Cursor and Claude Code have brought to software development.
“What agentic AI did for software, Empirik wants to do for infrastructure engineering,” he said.
The timing could give Empirik a broader opportunity as enterprises adopt AI agents capable of making autonomous changes to production systems. The challenge is that infrastructure failures can have consequences far beyond a flawed piece of code, including service interruptions, financial losses and disruptions to critical business operations.
That development has also created a high bar for autonomous infrastructure systems. Companies are likely to demand strong controls, auditability and human oversight before allowing AI agents to make consequential changes to production environments.
Empirik faces competition from AI-powered site reliability and observability platforms, including Resolve and Sequoia-backed Traversal. Balkansky says that Empirik occupies a distinct position by concentrating on understanding infrastructure changes and their potential downstream effects rather than simply responding to incidents after they occur.
Its central proposition is a shift from reactive observability to predictive infrastructure management. If Empirik can accurately determine which changes are likely to cause failures before they reach production, it could help companies reduce downtime while allowing engineering teams to manage complex technology environments with fewer manual interventions.