They met at 10 p.m. each weeknight, after class at Columbia University.
Sometimes talking over Zoom until 2 AM, Anish Agarwal, Raaz Dwivedi, Ahmed Lone, and Raj Agrawal talked about what it might mean to walk away from their lives—for a startup. All academics in some form or fashion, the group swapped ideas at the intersection of their research: causal machine learning, reinforcement learning, and AI agents.
For Agarwal, the decision wasn’t to be taken lightly. That semester, he’d just started a tenure-track faculty position at Columbia after earning his PhD from MIT—he had an academic career that was taking shape. But he was drawn to the entrepreneurial unknown.
“I had to get in the game,” said Agarwal, who’s originally from Singapore. “I was looking around at my peers—for example, at Cartesia and Reflection AI. They’re all super smart, and out here building companies. By DNA, they’re researchers with some commercial instinct. They’re solving really hard technical problems, and continuing to produce research in a very cool package. That felt right. That was my motivation.”
From those late-night meetings came Traversal, a startup founded in 2023 that focuses on observability and site reliability engineering (SRE)—helping engineers pinpoint and troubleshoot complex software failures with speed and precision. Troubleshooting is “one of the most complex workflows in software,” said Agarwal, Traversal’s CEO. “It’s why you can have 50 people firefighting in a war room until they find the answer.”
Today, Traversal launches from stealth with $48 million in funding from its seed and Series A rounds, Fortune has exclusively learned. Sequoia led the company’s seed round, while Kleiner Perkins led Traversal’s Series A. Nat Friedman and Daniel Gross’s NFDG and Hanabi (led by Mike Volpi and Bryan Offutt) are also investors. Traversal—named for both the computer science concept of a graph traversal and the idea of journeying through complex systems—counts among its customers Digital Ocean, Eventbrite, Cloudways, and a number of undisclosed Fortune 100 financial services companies. The company and others like it exist to stop software infrastructure crises and limit downtime.
“Imagine you’re having a heart attack right now,” said Agarwal. “That’s the only thing that matters. It doesn’t matter what happens ten minutes from now, or what your dinner is going to look like. An engineering team has two heart attacks a week, and a debilitating condition. So, you never get time to think about planning ahead and being great.”
The bigger the company, the higher the stakes of this problem: Bratin Saha, chief technology and product officer at DigitalOcean, said via email that “the sheer volume and complexity of our cloud infrastructure—serving hundreds of thousands of customers—means that even minor platform incidents can quickly escalate, impacting customer experience and incurring significant costs.” Saha told Fortune that, over six months, Traversal has helped resolve problems 37% faster. Traversal’s traction over the last two years is both a product of this moment in tech, the AI wave—and timeless.
“Whenever something new shows up, whether it’s a new trend, new market, new kind of product, or a new kind of technology, guess what? It needs to be monitored,” Sequoia partner Bogomil Balkansky said. “That’s observability. And it needs to be secured, which is security. And that’s why those two domains—observability and security—periodically produce big winners.”
Some winners from observability in recent years that Balkansky points out: Splunk, Datadog, Dynatrace, AppDynamics, and New Relic. For Traversal—which sits at the intersection of observability and AI, and builds on existing observability tools—this means there’s opportunity, especially given that the company’s riding the vibe coding wave.
“The amount of code being written was already growing at a staggering rate, and with AI code generation, it’s accelerating like never before,” said Kleiner Perkins partner Mamoon Hamid. “With more code created by AI, there is more surface area to troubleshoot. There is a need for AI to autonomously troubleshoot, mediate and even prevent complex incidents at scale—self-healing codegen.”
The academic-to-entrepreneur path that Agarwal and his cofounders now walk is well-trod, from VMware’s Mendel Rosenblum to Databricks’s Ali Ghodsi, and more. And though there’s much they don’t have in common, academic research and startups do share at least one key feature.
“Most jobs, you’re at A and you have to get to B, and getting from A to B is really hard,” said Agarwal. “In research, though, you don’t really know where A is and you don’t know where B is. And even if you do know, it’s still very hard to get to—and I loved that. I enjoyed the uncertainty.”
ICYMI…Here’s our exclusive on the New York-based health tech startup Tennr, which raised a $101 million Series C at a $605 million valuation to tackle the convoluted web of patient referrals.
Here we go again…Elon Musk’s xAI, according to Bloomberg, is in talks to raise $4.3 billion, adding to the $5 billion in debt he’s seeking for the company.
See you tomorrow,
Allie Garfinkle
X: @agarfinks
Email: alexandra.garfinkle@fortune.com
Submit a deal for the Term Sheet newsletter here.
Nina Ajemian curated the deals section of today’s newsletter. Subscribe here.
This story was originally featured on Fortune.com
