Faros AI is a key player in the management and observability layer of the AI agent stack. As organizations deploy AI agents for coding, testing, and deployment, Faros provides the infrastructure to measure the efficiency and quality of agent-generated work. Their platform includes specific extensions for tracking AI coding assistants and agents, helping teams understand the "human + agent" productivity loop.

The company is also pushing into the "knowledge" side of agents. By embedding an organization's specific engineering context and tribal knowledge into the data graph, they enable agents to write code that aligns better with internal standards and passes reviews faster. For people building or deploying agents at scale, Faros is the verification layer that ensures these agents are actually improving the engineering system rather than just increasing the volume of code without a corresponding increase in value.

The visibility gap in software delivery

For most enterprise departments, the transition to data-driven management happened a decade ago. Sales leaders have CRMs, and marketing teams have sophisticated web analytics. Engineering leadership, however, has historically operated in a relative black box. While individual tools like Jira or GitHub provide local data, they rarely offer a unified view of how code actually moves from a developer's machine to a production environment. Faros AI is built to close this gap by creating an engineering data graph that normalizes and connects these fragmented signals.

Founded in 2019 by Vitaly Gordon, Shubha Nabar, and Matthew Tovbin—all veterans of Salesforce's data and engineering leadership—the company is based in the San Francisco Bay Area. The founders' experience at Salesforce is central to the product's philosophy: managing thousands of engineers requires the same level of analytical rigor as managing thousands of customers. Faros is designed to provide this rigor, helping VPs of Engineering and CTOs move away from anecdotal evidence toward quantified performance.

Quantifying the AI coding transition

The current industry-wide push to integrate AI coding assistants has created a new challenge for engineering leaders. While tools like GitHub Copilot are widely adopted, many organizations struggle to measure their actual impact on productivity. Faros has positioned itself as the measurement layer for this transition. The platform tracks how AI tools influence DORA metrics—such as lead time for changes and deployment frequency—while also monitoring for potential risks like a decrease in code quality or an increase in technical debt.

By providing a specific "Copilot Module," Faros helps organizations identify where AI agents are underused and where they are effectively accelerating the development cycle. This level of granularity is intended to help executives make data-backed decisions about AI budgets, shifting the conversation from "is this tool cool?" to "what is the measurable ROI?"

Under the hood: The engineering data graph

Unlike simple dashboarding tools that sit on top of Jira, Faros AI uses a more complex data ingestion model. It connects to the entire software development lifecycle stack, including source control, CI/CD, and even communication tools. This data is normalized into a single schema, allowing for cross-tool analysis that would be difficult to perform manually. For example, a leader can see not just that a ticket is stuck, but that it is stuck because of a specific bottleneck in the automated testing pipeline or a delay in the code review process.

This system-level view is used by enterprise teams at companies like Autodesk, Coursera, and Vimeo. For these organizations, Faros is less about monitoring individual performance and more about identifying systemic blockers that prevent teams from being productive. The platform provides benchmarks that help teams compare their performance against industry standards, such as DORA's elite, high, medium, and low performance tiers. In a competitive market where engineering talent is expensive, the goal of Faros is to ensure that talent is spending its time on value-added work rather than navigating friction in the development process.