Sequoia-incubated Empirik launches with $21M to predict IT outages
The new observability startup uses agentic AI to track system dependencies and prevent infrastructure failures before they occur.
Empirik, a startup incubated by Sequoia Capital, has officially spun out as an independent company to tackle the growing complexity of IT infrastructure. The company aims to shift site reliability engineering (SRE) from reactive firefighting to predictive prevention.
The startup launched with $21 million in seed funding provided by Sequoia, Canapi, and Alumni Ventures. Led by CEO Kartik Chandrayana, who previously served as the observability VP at Salesforce and CPO at Quantum Metric, Empirik provides an AI-driven observability tool. The platform acts as an autonomous "traffic cop" for DevOps and SRE teams by tracking system changes and their subsequent ripple effects to predict and prevent outages. Early adoption is already underway, with a customer base that includes Guardant Health, S&P Global, and a major consumer packaged goods company.
The shift to predictive observability
Empirik was incubated by Sequoia in 2023, emerging from a realization among infrastructure veterans that large language models (LLMs) could fundamentally change IT operations. While traditional observability tools monitor system health, Sequoia partner Bogomil Balkansky noted that most existing tools fail to understand the complex system dependencies that often lead to cascading failures. By leveraging AI to map these dependencies, Empirik identifies risks that human operators or legacy monitoring systems might miss.
Why agentic AI matters for SREs
As AI accelerates the pace of software development, the sheer volume of system changes has increased, making infrastructure stability more difficult to maintain. Empirik is applying "agentic AI" to infrastructure engineering to automate routine troubleshooting and risk assessments. According to CEO Kartik Chandrayana, the goal is to do for infrastructure engineering what agentic AI has already done for software development, citing the transformative impact of tools like Cursor on coding.
By offloading the burden of routine maintenance and failure prediction to an autonomous layer, the company argues that SREs can move away from manual toil and focus on high-value architectural improvements. This approach positions the tool as a complementary layer to existing AI SRE platforms, such as Traversal and Resolve, rather than a direct replacement.
What to watch
As Empirik moves beyond its incubation phase, the industry will be watching whether its predictive models can significantly reduce mean time to recovery (MTTR) and prevent high-severity outages in complex enterprise environments. While the company has secured high-profile early customers, the primary challenge remains the ability of AI to accurately map dependencies across diverse, legacy, and hybrid-cloud stacks without generating excessive noise for engineering teams.