Mendral: Building the AI DevOps Engineer for CI/CD

Mendral positions itself not as another DevOps tool, but as an AI DevOps Engineer—a fundamentally different category in the software delivery ecosystem. Founded in 2025 and launched as part of Y Combinator’s Winter 2026 batch, Mendral is built to observe, diagnose, act, and learn continuously across CI/CD pipelines. Instead of generating alerts or dashboards that still require human intervention, Mendral takes responsibility for fixing the problems that slow teams down.

At its core, Mendral is designed to autonomously manage delivery reliability. It handles CI failures, flaky tests, slow builds, broken releases, and even aspects of code review—improving with every build and run. Over time, it evolves into a self-improving delivery system that manages performance, quality, security, and compliance without constant manual effort from engineers.

In an industry where DevOps complexity keeps growing, Mendral proposes a bold idea: what if CI/CD didn’t need to be actively managed by humans at all?

Why Is CI/CD Still a Major Pain Point for Modern Engineering Teams?

Despite years of tooling innovation, CI/CD remains one of the most frustrating parts of software development. Engineering teams lose thousands of hours each year chasing broken builds, flaky tests, and fragile pipelines. Traditional DevOps tools are excellent at collecting data, emitting alerts, and visualizing failures—but they largely stop there.

The burden of understanding why something broke, what changed, and how to fix it still falls on developers and platform engineers. As teams grow, this often leads to the creation of dedicated DevOps or platform roles whose primary responsibility is keeping delivery systems functional rather than delivering business value.

The situation is getting worse, not better. Coding agents and AI-assisted development have dramatically increased code output, placing even more load on CI systems. More commits mean more builds, more tests, and more opportunities for failures. Yet the underlying CI infrastructure and workflows were not designed for this scale or complexity.

Mendral was created in response to this mismatch—where modern development speed collides with outdated delivery operations.

How Does Mendral Rethink the Role of DevOps in the AI Era?

Mendral challenges the assumption that DevOps must always be human-driven. Instead of treating CI/CD as a system that engineers must constantly supervise, Mendral treats it as a domain where AI can operate autonomously.

The platform continuously observes signals across the delivery pipeline—build logs, test results, code changes, historical failure patterns, and performance metrics. From there, it diagnoses what went wrong, determines likely root causes, and takes action. Crucially, it doesn’t just execute static rules; it learns from outcomes, improving its future decisions.

This loop—observe, diagnose, act, learn—is what allows Mendral to behave less like a tool and more like an engineer embedded within the team. Over time, it builds an understanding of a team’s codebase, testing strategy, and delivery patterns, adapting to their unique environment.

Rather than replacing developers, Mendral removes the cognitive and operational overhead that distracts them from building product.

What Can Mendral Do Today Inside a CI/CD Pipeline?

Today, Mendral already operates in production CI/CD environments for 15 teams, with five paying customers, including PostHog. Its current capabilities focus on the most painful and time-consuming parts of delivery reliability.

Mendral can diagnose CI failures by analyzing what changed, what failed, and why. Instead of forcing developers to dig through logs and diffs, it provides actionable explanations tied directly to recent code or configuration changes.

It also detects and mitigates flaky tests—one of the most notorious productivity killers in software teams. By identifying patterns across runs, Mendral can flag unreliable tests and help teams stabilize their pipelines without endless trial and error.

Slow builds and test runs are another major focus. Mendral identifies performance bottlenecks and proposes fixes, helping teams keep feedback loops fast even as their codebases grow.

Additionally, Mendral supports code reviews with a focus on delivery reliability, surfacing issues that are likely to impact CI stability rather than just code style or correctness.

How Does Mendral Become a Self-Improving Delivery System Over Time?

What differentiates Mendral from automation scripts or rule-based tools is its ability to learn continuously. Every build, failure, fix, and outcome feeds back into the system.

As Mendral observes more runs, it builds a contextual understanding of which failures are meaningful, which tests are historically flaky, and which changes tend to introduce instability. This allows it to move beyond reactive fixes and toward proactive prevention.

Over time, Mendral expands its scope from fixing individual CI issues to managing broader delivery concerns such as performance, quality, security, and compliance. Instead of treating these as separate tools or workflows, Mendral integrates them into a unified delivery intelligence layer.

For smaller teams in particular, this means gaining enterprise-grade delivery reliability without needing a dedicated platform engineering function.

Who Is Building Mendral and Why Does Their Background Matter?

Mendral is founded by Sam Alba and Andrea Luzzardi, two engineers with deep roots in infrastructure and developer tooling. Both founders were early engineers at Docker, where they helped shape some of the most influential tools in modern software development. Andrea wrote Docker’s first lines of code, while Sam became Docker’s first hire and later served as VP of Engineering.

Together, they went on to co-found Dagger, a programmable CI/CD engine that reimagined how pipelines could be defined and executed. Their experience spans startups and large organizations alike, including time at Google and Microsoft.

This background matters because Mendral is not built from theory—it is built from firsthand experience with CI pain at scale. The founders understand how fragile pipelines become as teams grow and how costly manual DevOps operations can be.

Mendral reflects a long-term vision shaped by years of working directly on the infrastructure that underpins modern software delivery.

Why Is Mendral Especially Relevant in a World of AI Coding Agents?

AI coding agents are accelerating software development at an unprecedented pace. While this promises faster feature delivery, it also amplifies stress on CI/CD systems. More generated code means more builds, more tests, and more potential failure modes.

Without automation on the delivery side, teams risk trading development velocity for operational chaos. Mendral addresses this imbalance by applying AI to the other half of the equation—delivery reliability.

By autonomously managing CI/CD complexity, Mendral ensures that faster code production does not result in slower releases or lower quality. It acts as a stabilizing force, allowing teams to scale their use of AI coding tools without scaling their DevOps burden.

In this sense, Mendral is not just complementary to AI development—it is essential infrastructure for making AI-accelerated engineering sustainable.

How Does Mendral Reduce the Need for Dedicated Platform Engineering Teams?

Many small and mid-sized teams struggle with a difficult trade-off: invest in dedicated platform engineers or accept unreliable delivery pipelines. Mendral offers a third option.

By autonomously handling CI/CD reliability, Mendral reduces the operational load that typically justifies specialized DevOps roles. Teams can maintain fast, stable delivery pipelines without pulling senior engineers away from product work.

This does not eliminate the need for platform engineering in large organizations, but it dramatically lowers the barrier to entry for teams that want professional-grade delivery without enterprise-level overhead.

For startups and fast-moving teams, this can translate directly into faster iteration, lower costs, and better developer experience.

What Does Mendral’s Early Traction Say About Its Potential?

Managing CI/CD in production for 15 teams—and converting a third of them into paying customers—signals strong early validation. The fact that customers like PostHog are already using Mendral suggests that its value is not hypothetical.

Early traction demonstrates that teams are willing to trust an autonomous system with critical parts of their delivery pipeline. It also shows that Mendral is solving real, urgent problems rather than offering incremental improvements to existing workflows.

As Mendral continues to learn from real-world usage, its capabilities and reliability are likely to compound, strengthening its position as a core part of modern engineering stacks.

What Is the Long-Term Vision for Mendral?

Mendral’s long-term ambition is to become a fully autonomous delivery system—one that continuously improves software performance, quality, security, and compliance without manual intervention.

Rather than adding yet another tool to the DevOps landscape, Mendral aims to absorb complexity and return simplicity to engineering teams. The ultimate goal is a world where developers rarely think about CI/CD unless they choose to.

If successful, Mendral could redefine how teams think about DevOps entirely—not as a discipline that requires constant attention, but as a background capability handled by intelligent systems.

In a future shaped by AI-driven development, Mendral positions itself as the AI engineer that keeps everything running smoothly, quietly, and reliably in the background.