Traversal Raises $120M Series B for AI Site Reliability

A customer who buys incident diagnosis can come back with a question about tomorrow's deployment. That widening relationship sits behind Traversal, the New York company building AI for site reliability engineering, and its ambition to take on more of the work surrounding live software.

Traversal announced a $120M Series B at a $1B valuation on November 3, 2026. Kleiner Perkins, Redpoint Ventures and Khosla Ventures co-led the financing, with Sequoia Capital, Battery Ventures and Amex Ventures participating, according to the company's announcements. The money supports research, engineering and go-to-market hiring as Traversal develops toward what it calls self-driving production.

The investment reaches into an awkward part of the AI software economy. Faster code creation puts more changes within reach, while the people operating those systems still need to understand what a change will do and why a failure occurred. Traversal is selling an understanding of production that can remain useful after the immediate incident has ended.

What Traversal's Series B adds to its funding history

The latest round brings Traversal's total funding to $173M, according to its issuer release distributed by Business Wire. That history includes the combined $48M seed and Series A funding disclosed when the company emerged from stealth in June 2025, followed by a strategic investment from Amex Ventures disclosed in December 2025. The $48M belongs to those earlier rounds together; treating it as the Series A alone would distort the financing record.

Sequoia's original investment account describes a long-standing interest in troubleshooting and the team's route into its Arc company-building program. The November financing adds Redpoint and Khosla as co-leads alongside returning backer Kleiner Perkins. For a young infrastructure vendor, that syndicate can support a business whose research and enterprise deployment work need to advance together.

There is an important commercial distinction inside that work. An incident gives an engineering team an immediate reason to evaluate a tool, but the evaluation happens in the customer's own environment, with its own dependencies and operational habits. Traversal's opportunity grows when the result is useful enough for that customer to bring another question. The product relationship can then extend beyond the emergency that first justified the purchase.

Why causal reasoning matters during a production failure

Traversal's founders are CEO Anish Agarwal, CTO Raaz Dwivedi, Raj Agrawal and Ahmed Lone. The company's founder biographies describe a combination of causal machine learning research, computer science and operating experience, including Lone's earlier work at Citadel Securities. That background matters because the problem is as much about explaining a chain of events as it is about finding unusual data.

Traversal describes its Production World Model as a changing representation of applications, services, infrastructure, dependencies, code and operational context. Its Causal Search Engine works through candidate explanations across those relationships. The aim is to identify a cause supported by timing and system behavior, rather than stop at a plausible story about signals that appeared together.

For an operator, the difference becomes practical quickly. A symptom can appear several dependencies away from the change that caused it. Acting on the visible symptom may consume time while the original problem continues, and adding an articulate AI explanation does little good if the underlying diagnosis is wrong. This is why access to the actual production environment matters to the buying decision: the customer needs the tool to understand the system that is failing.

The architecture also creates a continuing maintenance assignment. Services, configurations and dependencies change after deployment, so a useful picture of production has to change with them. Traversal's claim is that its model keeps that context current, giving an investigation somewhere more useful to begin than a collection of disconnected alerts.

How incident diagnosis becomes a broader operating relationship

Traversal says roughly two-thirds of its usage now extends beyond root-cause analysis, including alert triage, release risk and production support. That is the most revealing expansion claim in the announcement: customers are applying an understanding developed for incidents to other decisions about the same environment. The claim comes from Traversal, and its public materials do not establish the revenue contribution of each use case.

The company's customer evidence names Nubank, Harvey, PepsiCo and DigitalOcean. Those relationships put the product in settings with very different software and operating requirements. They also make clear why an enterprise customer might care about evidence, access controls and deployment choices as much as the speed of an answer.

Traversal says its platform deploys read-only by default and supports bring-your-own-cloud and bring-your-own-model arrangements. That gives a buyer a way to begin with observation and investigation while deciding which additional responsibilities to allow. A broader autonomous workflow introduces further questions about authorization, review, corrective action and verification that an initial diagnosis alone cannot settle.

DevCuration's coverage of &Truth's software verification work and Larridin's AI measurement platform follows related questions about how enterprises evaluate AI-generated work. For production reliability, the buyer has several possible routes: continue with existing observability tools and human investigation, assemble an internal agent, or adopt a specialized AI SRE product. Traversal's proposition depends on how well its accumulated production context travels between those tasks. The convenience of another interface will carry less weight than the confidence an engineer can place in the answer during an actual failure.

Where the new capital takes the production handoff

Traversal Workers can initiate investigations proactively, and the company's MCP integration can provide production context to coding agents. Together, those capabilities connect what engineers learn while operating software to the next changes they make. The broader self-driving-production ambition includes detecting problems, determining causes, taking corrective action and checking the result.

Building that relationship requires an organization around the research. Traversal's leadership expansion brought in Maxime Petazzoni as Head of Engineering alongside senior sales, marketing and field-engineering hires. Enterprise adoption requires the vendor to understand how a customer deploys, evaluates and governs the system, then carry that knowledge into ongoing delivery.

The Series B funds more of that work while customers decide how much operational authority to extend. Inside each deployment, an engineer will still need to understand why Traversal recommends a change, which services it may affect and how the team will know the intervention helped. As the product moves further into production, those decisions become part of the customer relationship Traversal is asking to keep.