Why Canapi is Investing in Empirik

Every Outage Has an Author

Ask any infrastructure team what caused their last serious incident and the answer is almost never mysterious, but simultaneously somewhat unclear. Someone attached a policy. Someone flipped a flag. Someone merged a Terraform change on a Friday afternoon. The failure was authored, and it was authored recently. The difficulty is not that the cause is unknowable; it is that proving it takes hours of manual correlation across a dozen systems that were never designed to talk to one another.

Enterprises have spent enormously to shorten those hours. The observability market is a mature, multi-billion-dollar category, and it is very good at the job it was built for: telling you that latency has spiked, that a pod has crashed, that error rates are climbing. What it does not tell you is who did it and what else they touched. While the alert is the smoke, it is the change that is the fire.

And that fire is suddenly spreading at a more rapid rate than ever before, and of course, it is thanks to AI. Coding assistants have compressed the time it takes to produce software from weeks to hours, which means the volume of things arriving at the infrastructure layer has multiplied: more deploys, more permissions granted, more services spun up, more dependencies quietly rearranged. The systems governing all of it, however, still run at the speed of a human reading a ticket. Change advisory boards, approval queues, and spreadsheets do not scale to meet machine-generated change, and adding headcount does not fix an architectural problem.

The category built to solve this challenge has not delivered. Global spend on IT service management reached roughly $11.9 billion in 2024 and is projected to approach $37 billion by 2032, yet Gartner estimates only about a quarter of organizations extract meaningful value from their configuration management database investments, chiefly because the data inside them is wrong. Anyone who has watched a team export thousands of change tickets to a CSV in order to feed a CMDB understands why. Ask most large enterprises what is currently running in their environment and you will get an answer that was accurate the last time somebody updated it by hand.

That’s why we’re so excited about Empirik.

Putting Change to Work

Empirik treats change as a first-class signal rather than an administrative artifact. Alongside the three established pillars of observability, the metrics, logs, and traces that describe symptoms, Empirik captures the change events themselves and makes them queryable.

Mechanically, the platform pulls metadata from a customer's clouds, identity providers, CI/CD pipelines, Kubernetes clusters, SaaS applications, and on-premises systems, and resolves all of it into one live repository of assets, accounts, and the relationships between them. Code has had repositories for years, but infrastructure never has – until now. That distinction is what enables the useful behavior: because Empirik knows how the pieces connect, it can evaluate a proposed change at the moment it originates and calculate what that change will reach before anyone approves it. Risk scoring, affected owners, downstream blast radius, and policy conflicts all become answerable in advance rather than reconstructed during a postmortem.

Most tools in this space assemble their picture from network topology, which reveals what is talking to what. Empirik builds from change history instead, which reveals something more useful: who and what has authority over each resource, and therefore who needs to know when it moves.

Empirik emerges from stealth with a world-class team. Founded from within Sequoia, Empirik is the brainchild of Avon Puri, Chief Digital and Information Officer at the firm, who previously ran infrastructure as SVP and CIO at Rubrik and led engineering and applications at VMware, and Sudheer Dhurjati, who worked alongside him across all three organizations. Kartik Chandrayana joined as CEO earlier this year; he founded and sold Twin Prime to Salesforce, spent five years there as VP of Product for observability and big data, and most recently served as Chief Product Officer at Quantum Metric. It is an unusually well-matched combination for this problem: two operators who lived the pain from inside the enterprise IT seat, and a product leader who has shipped observability at scale to the buyers Empirik is now selling to.

Why This Belongs in the Canapi Portfolio

Unauthorized and poorly understood change is a particularly expensive problem in banking. Financial institutions run some of the most tangled application estates in the economy, and they run them under examination. A mid-sized bank may depend on several hundred SaaS applications; the largest manage well over ten thousand. In that environment, the ability to flag a risky change before it merges, to detect drift between what was approved and what is actually running, and to produce an audit trail showing what happened rather than what was requested is not an efficiency gain. It is a control.

The early results support the thesis. Empirik is processing more than a million change and telemetry events daily across dependency graphs covering upward of 100,000 resources. Customers are reporting 40% to 70% reductions in incident triage time and 60% faster attribution to the change responsible for an incident. A payments processor handling over $1.5 billion in annual volume now catches sensitive configuration changes in ten seconds, down from thirty minutes. A managed service provider wired Empirik into its GitHub Actions to block risky changes at the source. Another large healthcare customer expanded from production into every development environment and then on-premises. A top global consumer goods company is building much of its autonomous infrastructure roadmap on the platform.

For years, the honest answer to "what changed?" has been a war room, a shared screen, and several people guessing in parallel. Empirik is building the system of record that answers it, first for the engineers doing the work today and then for the agents that will increasingly be doing it tomorrow. We are thrilled to back Kartik, Avon, Sudheer, and the Empirik team as they bring infrastructure operations into the autonomous era.