- Kodesage has raised $6.6 million in a seed round led by VentureFriends, with pre-seed backer Portfolion returning.
- The London and Budapest-based startup helps heavily regulated enterprises modernise decades-old software entirely on-premise, without sensitive code ever reaching public cloud AI tools.
- Backers include xAI co-founder Christian Szegedy and German footballer and World Cup winner Mario Götze, whose investment vehicle Companion M has backed more than 70 companies.
The most important software in a bank is often the software nobody fully understands anymore. It runs core processes, has run them for decades, and the engineers who wrote it have long since retired. Changing it is slow, expensive, and risky, which is why so much of it never changes at all.
London and Budapest-based Kodesage believes it has found a way to bring generative AI to exactly this problem. The startup helps heavily regulated enterprises modernise decades-old software systems safely on-premise, without exposing sensitive data to public cloud AI tools.
Kodesage has raised $6.6 million in a seed round led by VentureFriends, the Athens-based VC that has also backed industrial AI startup Sybilion with pre-seed lead Portfolion returning. Angels joining the round include xAI co-founder Christian Szegedy and German footballer and World Cup winner Mario Götze, whose investment vehicle Companion M now holds a portfolio of more than 70 companies, two of which became unicorns in 2025. The round follows a €2.3 million pre-seed raised in early 2025, led by Portfolion.
The fresh funding will accelerate Kodesage’s go-to-market across the US and Europe, and will power continued investment in engineering and product to deliver on the company’s vision of self-healing enterprise applications.
Born from a problem the founders experienced firsthand
Kodesage was founded in 2024 by Gergely Dombi, Miklos Szurdi, and Gyorgy Szilagyi. Dombi and Szurdi previously built Sonrisa, a software consultancy that grew to more than 300 employees specialising in enterprise system modernisation. They repeatedly encountered the same roadblock: legacy transformation projects that dragged on for years because institutional knowledge lived entirely inside a handful of retiring specialists’ heads. Szilagyi is best known as co-founder of Tresorit, the encrypted cloud storage platform acquired by Swiss Post in 2021.
“Software modernisation is rarely clean cut. In regulated environments, legacy and new systems often coexist for years, whilst the support burden grows as institutional knowledge thins out. This is a multi-trillion-dollar drag on enterprise IT. Kodesage’s living knowledge layer helps teams modernise legacy systems faster, resolve incidents and reduce repeat issues through powerful AI-assisted support.
“Our vision is self-healing enterprise applications: systems that can continuously learn, propose, test, and validate fixes, with engineers guiding outcomes rather than diagnosing problems and writing code from scratch. For regulated enterprises whose support teams have been under growing operational pressure for years, this is where the deepest value lies.” — Gergely Dombi, co-founder and CEO, Kodesage.
Why legacy software remains a hard problem for AI
Modern AI coding assistants thrive on clean, cloud-native codebases. They fundamentally struggle with legacy architectures written in COBOL, PL/SQL, PowerBuilder, or Oracle Forms — languages where business logic sits buried in stored procedures, database schemas, and decades of undocumented configuration. Heavily regulated organisations also cannot allow sensitive source code or proprietary database schemas to leave their perimeter to hit public AI clouds.
Kodesage operates entirely inside the customer’s secure environment — on-premise, inside a virtual private cloud, or within fully air-gapped infrastructure. The platform performs automated deep discovery to extract buried logic and fuses it into a living knowledge layer that powers documentation, context-aware code conversion, automated test generation, and production support. The company says its platform can reduce manual documentation workloads by more than 80% and accelerate migrations by up to three times. The approach puts Kodesage in a fast-growing category: London’s AppFactor raised $4M earlier this year to automate enterprise software maintenance with AI agents, targeting the same $1.5 trillion technical debt problem.
The shift from pitch to portfolio
Professional athletes are increasingly moving beyond real estate to back tech startups. As TFN has reported, Lionel Messi launched his own Silicon Valley investment vehicle, while Olivier Giroud and N’Golo Kanté backed Upway’s $25M Series A. For Götze, Kodesage fits a deliberate pattern his Companion M vehicle focuses on early-stage software, healthcare, and cybersecurity, with ticket sizes typically between €25,000 and €50,000.
“There are numerous enterprises globally struggling to maintain and upgrade legacy software systems that reside on-premise. Starting with Oracle applications, Kodesage can help companies understand, maintain, and modernise complex, undocumented legacy codebases. As a result, they save them time and cost while cutting their dependence on retiring experts. Gergely, Gyorgy and Miklos have deep experience in the space, and we couldn’t be more excited to back them in their journey.” — Apostolos Apostolakis, Founding Partner, VentureFriends.
The competition
Kodesage competes in a crowded but fragmented space. Stardog Union, which has raised around $25 million to help large organisations navigate complex data webs using enterprise knowledge graphs, focuses on data unification rather than legacy codebase modernisation. Cognition, which raised $1B at a $26B valuation, recently demonstrated its AI agent modernising a Mercedes-Benz legacy system in eight days — work that previously took eight months. Where Cognition targets cloud-native enterprise development broadly, Kodesage’s differentiator is the air-gapped, on-premise constraint that excludes most mainstream AI coding tools from regulated industries entirely.
Market context
According to Research and Markets, the legacy modernisation market was valued at $22.17 billion in 2025 and is projected to reach $46.5 billion by 2030, growing at a CAGR of 15.9%. The pipeline of demand is structural: an estimated 80% of Fortune 500 companies still run active legacy systems, and the pool of engineers who understand them continues to shrink as that generation retires.
The bigger question for Kodesage is not whether enterprises want their legacy systems modernised, every CIO does. The question is whether a $6.6 million seed-stage startup can move fast enough before the cloud-native giants decide this problem, with its air-gapped constraints and all, is finally worth solving on their terms.
