We’ve been looking for a company like Traversal for years. 

For more than a decade, we at Sequoia have been obsessed with “the rise of the developer,” guiding our partnerships with companies that cater to engineers like GitHub, MongoDB and Stripe. And Bogomil has been fascinated with observability for years: shortly after joining Sequoia in 2020, after a friend pointed out to him that “observability is a s***show,” he identified that a new generation of tools would help technical teams care for their systems.

More recently, it became clear that the road to reinventing observability ran through AI. Back in 2023, as GitHub Copilot caught lightning in a bottle, Charlie and our partners Stephanie and Josephine recognized that founders might apply AI to many problems engineering teams face beyond writing code. One we were excited about in particular was troubleshooting, the painstaking process of finding the root cause of a bug, often using observability data.

All we needed to do was find the right team to work with. 

Believe it or not, our relationship with Traversal’s team dates back years, too. Traversal’s founder, Anish Agarwal, knew our partner, Shaun Maguire, and former partner Dan Chen through their shared alma mater, Caltech. They remembered him as the star student who somehow managed to triple major, excel on the tennis team and still have a social life. Charlie knew co-founder Ahmed Lone, met the rest of the team and encouraged them to apply to Arc, our company-building program for Sequoia pre-seed and seed-stage founders.

They were the type of team we were looking for. In our view, AI troubleshooting was first and foremost an AI problem, and an observability problem only second. Traversal’s team of three Ph.D.s from MIT and Berkeley, who’d studied causal ML and reinforcement learning—and one former Citadel Securities quant trader who’d lived this problem firsthand—seemed like the right fit. 

In March 2024, when we finally all got in one room—Sequoia’s San Francisco space atop a chocolate factory—it was obvious to us as investors what to do. 

Observability is one of the crown jewels of infrastructure and a hotspot for generational companies. Businesses like Splunk, Datadog, Grafana and Cribl help thousands of engineers monitor their systems and generate billions in annual revenue in the process. And still, the job is unfinished. Troubleshooting remains deeply painful.

Picture getting woken up at 3 a.m., gathering with dozens of people in a Slack war room and “dumpster diving” through heaps of observability data to find the proverbial “needle in a haystack” root cause. That’s what engineers today do to troubleshoot. It takes hours, sometimes days, sometimes the needle is never found at all. When there are issues (and every software company has issues), slow troubleshooting costs companies dearly in terms of missed revenue and customer trust. And it costs devops and site reliability engineers (SREs) dearly in terms of lost sleep and missed dates.

Troubleshooting is so painful, in fact, that “Every engineer remembers their first time”—their first production incident, that is—in the words of Traversal (then Interaction Labs) co-founder Ahmed, a former quant trader at Citadel Securities, where troubleshooting is about as high stakes as it gets. Below, he talks about a harrowing experience of troubleshooting, with minutes before the market opened. 

Traversal automates troubleshooting. Their AI SRE uses a unique mix of LLMs and causal ML to do the dumpster diving, finding root causes automatically. Traversal does the dirty work, so engineers can fix problems ASAP. Companies save money and gain customer trust, and engineers no longer need to pull their hair out in the middle of the night. 

Traversal only got started a year ago, and already they’re showing they can deliver real business ROI to companies like DigitalOcean, American Express, Cloudways, Eventbrite and more. LLMs aren’t always great troubleshooters out of the box, but Traversal’s team has developed a unique approach: swarms of parallel investigations with proprietary statistical tools make them immediately effective at enterprise scale. 

Combine their technical skills and sense of humor with their customer obsession and ability to recruit exceptional talent, and you’re starting to get a picture of Traversal’s team. That’s why we’re so excited to be their partners after leading their seed round a year ago, and to support their mission: making life for engineers and engineering teams much, much better.