A $21M seed bets that most downtime starts with a change nobody traced properly.
Most observability tools tell you what broke. Empirik is built to tell you what is about to. The startup, incubated inside Sequoia Capital in 2023, spun out as an independent company Tuesday with $21 million in seed funding from Sequoia, Canapi, and Alumni Ventures. The pitch: track every system change, then infer where it will ripple.
Where the idea came from
Avon Puri joined Sequoia as chief digital and information officer in 2020, after more than a decade running infrastructure at Rubrik and VMware. Around three years ago, watching large language models mature, he and fellow Sequoia IT leader Sudheer Dhurjati landed on a specific question. Infrastructure teams spend their lives reacting to failures. Could AI get ahead of them instead?
That flipped the problem. Rather than diagnosing an outage after the pager goes off, the product maps changes across an environment and works out what each one touches. Sequoia saw a new category forming and backed it. Earlier this year the firm brought in Kartik Chandrayana — formerly CPO at Quantum Metric and observability VP at Salesforce — as CEO.
How it actually behaves
Sequoia partner Bogomil Balkansky describes it as an autonomous traffic cop sitting over the change stream:
- Low risk — waved through without human involvement
- Larger changes — allowed, but with guardrails attached
- Dangerous updates — flagged and escalated for human review
His argument is that money has always flowed into keeping systems running, but existing tools do not really understand how complex dependencies connect. Empirik, he says, is among the first products aimed squarely at tracking change across sprawling environments.
The practical effect is offloading. DevOps and site reliability teams hand over routine troubleshooting and spend their hours on work that matters more. Chandrayana frames the ambition plainly: what agentic AI did for software, Empirik wants to do for infrastructure engineering. The internal comparison is Cursor and Claude Code — tools that made developers meaningfully faster by automating the tedious parts.
Customers are already on board. The range runs from startups to Fortune 500 names including S&P Global and Guardant Health, plus a large consumer packaged goods company. That is unusual traction for a company that only launched this year.
On competition, Balkansky claims the category is currently empty. He positions Empirik as a complementary layer above AI SRE platforms such as Resolve and Sequoia-backed Traversal, not a replacement for them.
AI is pushing code into production faster than change-management processes were designed to handle. If your engineering org has quietly accepted more deploys and more incidents as a fair trade, this is the decision to revisit. Prevention tooling is becoming a real budget line, separate from monitoring. Ask what your current stack actually knows about dependencies before renewing it.
