Today, Marathon VC is leading a €1.1 million Seed round for Straion, the rules layer built to turn the current chaos of AI coding into governed, production-grade engineering.
We have entered the “Prompt-and-Pray” era of software development.
Tools like GitHub Copilot and Cursor have turned every developer into a high-speed generator, but for engineering leaders at scale, this velocity is creating a new kind of crisis. We are seeing more code than ever before, but it often lacks the “organisational DNA” required to survive in a complex enterprise environment.
The current struggle with AI-first coding is the constant need for manual course correction. An AI agent might suggest a brilliant piece of logic that technically works, but it doesn’t know your company’s specific Kafka naming conventions or your PII masking protocols. This leads to an exhausting trial-and-error loop where senior engineers spend their days “babysitting” the AI’s work to ensure it doesn’t break architectural patterns.
Straion helps teams centralise engineering standards in a single rule hub, dynamically select the right rules for each task, and validate plans before implementation — not just after code generation. It integrates seamlessly with existing workflows such as Claude Code, Cursor, and Copilot.
The goal is simple: move faster, reduce drift, and increase confidence in the reliability of generated code.
“The industry has spent the last two years obsessed with making AI faster. But in an enterprise environment, speed without alignment is a liability,” says Lukas Holzer, co-founder of Straion.
“We built Straion to give AI the organisational context it was missing—moving it from a trial-and-error tool to a precision instrument that understands how your company actually builds software.”
Straion solves this by transforming static documentation into active, machine-readable guardrails. It doesn’t just wait for a mistake; it provides the AI with the right context at the right millisecond. Most importantly, it validates the AI’s plan before implementation begins.
This shifts the process from reactive “cleanup” to proactive precision. By building a platform that uses machine learning to dynamically retrieve only the rules relevant to a specific task, they have enabled true, governed autonomy. It’s about giving the AI the steering wheel it was missing.
According to Marathon VC, to understand why Straion is the missing piece of the modern dev stack, you have to look at the founders’ roots in Linz, Austria.
This isn’t a group of “vibe coders” chasing a trend; they are seasoned operators who have spent a decade in the trenches of enterprise software.
Lukas Holzer, Fabian Friedl, and Katrin Freihofner were colleagues at Dynatrace, the observability giant. While at Dynatrace, they noticed a recurring friction point: as teams grew, the “invisible rules” of the organisation—architectural standards, security mandates, and naming conventions—became increasingly difficult to enforce. These rules usually lived in “rotting” documentation: Confluence pages that no one read and 300-page PDFs that were updated once a year.
When AI agents began generating code at superhuman speeds, this “documentation gap” became a canyon. The AI could write a function in milliseconds, but it had no idea how that function fit into the broader organisational architecture.
“Most investors are looking for the next AI code generator. We were looking for the guardrails,” says Panos Papadopoulos, Partner at Marathon VC.
“Lukas, Fabian, and Katrin aren’t just building a tool; they are building the governance layer that makes the autonomous future possible for the enterprise.
They have the technical pedigree from Dynatrace to solve what we believe is the most critical bottleneck in modern engineering. We are proud to back this team as they build the governance layer that will define the next decade of software engineering.”
The funding will accelerate three priorities: deepening the product’s capabilities in rule governance and plan-stage validation, expanding integrations for scaled engineering workflows, and hiring mission-driven builders across AI engineering and full-stack development.
