Kodesage Raises $6.6M Seed to Modernize Legacy Software with On-Premise AI

A Founding Team Built for the Problem

Kodesage is led by a team with deep, practical experience in the problem they are solving. CEO Gergely Dombi and team member Miklos Szurdi previously founded and scaled a more than 300-person consultancy dedicated to modernizing legacy software systems. Gyorgy Szilagyi, the third co-founder, was previously a co-creator of Tresorit, the zero-knowledge, encrypted cloud storage company acquired by Swiss Post .

This combination of technical and security expertise directly informs the company's core technical decisions. The team operates out of London and Budapest .

An On-Premise AI Engine for Undocumented Logic

The Kodesage platform focuses on an automated process the company calls "deep discovery." It is designed to ingest legacy codebases—specifically naming COBOL, PL/SQL, PowerBuilder, RPG, and Oracle Forms—and map the embedded business rules, many of which are not formally documented .

This process generates a "living documentation layer" that is readable by both human development teams and AI agents. With this context established, the platform supports automated code conversion, generates tests, and provides what Kodesage calls AI-assisted production support .

The company's primary wedge market is Oracle Forms. It has developed a dedicated modernization kit, such as a path from Oracle Forms to Oracle APEX, which it claims can deliver conversions up to three times faster while reducing documentation effort by more than 80% .

Kodesage's long-term product vision is for "self-healing enterprise applications," where AI agents are able to automatically prepare, test, and implement routine fixes, with the resulting changes still subject to review by human engineers .

Deployment: A Firewall-First Approach

Unlike the vast majority of AI coding tools, Kodesage operates entirely on-premise. The platform is designed to run within a customer’s own environment, in a virtual private cloud, or in a fully air-gapped setting with no external network connection .

This means the raw source code, database schemas, and all other sensitive content never have to leave the organization's direct control. This architectural decision was heavily influenced by co-founder Gyorgy Szilagyi's prior work building Tresorit's zero-knowledge security model .

A direct consequence of this model is that Kodesage's pricing is detached from the per-token consumption pricing common among cloud-based AI services, which the company positions as more predictable and stable for enterprise budgets .

Target Industries and Expansion Plans

This "bring the AI to the data" architecture is squarely aimed at banks, insurance companies, energy and transport operators, and public sector agencies—any regulated institution that is prohibited from connecting its core database infrastructure to a public cloud .

With the new funding, Kodesage plans to use the capital for a formal go-to-market campaign in the United States and Europe, along with expanding its engineering and product teams .