Kodesage Raises $6.6M for On-Premise AI Legacy Modernization

The most critical software in global finance and insurance is often the software nobody fully understands anymore. It runs the core processes that move trillions of dollars, it has managed those processes for decades, and the engineers who originally wrote it have long since retired. Changing it remains a slow, expensive, and deeply risky endeavor, which is why so much of it never changes at all. A new startup called Kodesage is attempting to solve this entrenched problem with a freshly funded artificial intelligence platform designed to operate entirely within the secure boundaries of enterprise networks.

Kodesage secures $6.6 million in seed funding to advance its on-premise artificial intelligence platform for enterprise legacy modernization. The London and Budapest-based company targets regulated industries by keeping sensitive data within organizational boundaries while automating the discovery, documentation, and conversion of outdated core systems.

The Architecture of Institutional Memory

Enterprise software has evolved through distinct technological generations, each layer building upon the previous one without fully replacing it. The foundational systems that power banking and insurance operations were largely constructed during an era when computational resources were scarce. Programming languages prioritized raw efficiency over architectural elegance, establishing business rules directly into the codebase. Languages like COBOL and PL/SQL dominated this landscape, creating systems that managed daily transactions with remarkable reliability. Over time, these systems accumulated decades of incremental patches, regulatory adjustments, and operational workarounds. The original architects have retired, and the institutional knowledge required to navigate these complex codebases now resides in the minds of a shrinking group of specialists.

When those specialists leave, the organization faces a critical vulnerability. The software continues to function, but the ability to modify it safely diminishes rapidly. This creates a paradox where the most vital systems become the most difficult to update. Institutions are forced to rely on fragile patchwork solutions rather than comprehensive modernization efforts. The accumulated technical debt creates a barrier that discourages fundamental architectural changes. Organizations must constantly weigh the risk of disruption against the necessity of improvement. This dynamic has defined enterprise software management for decades, leaving many core platforms in a state of suspended animation.

Previous attempts to address this challenge relied heavily on human-led consultancy models. Teams of engineers would spend years manually tracing data flows, reverse engineering database schemas, and reconstructing business logic through extensive interviews and documentation reviews. While these projects occasionally succeeded, they consistently suffered from diminishing returns. The sheer volume of undocumented dependencies made comprehensive mapping nearly impossible within standard budget cycles. Organizations frequently abandoned modernization initiatives because the timeline stretched beyond corporate planning horizons. The fundamental issue was not a lack of technical capability, but rather an inability to scale human expertise across massive, interconnected codebases. The industry required a method to extract institutional memory at machine speed while preserving contextual relationships that human analysts struggled to map manually.

Why does on-premise deployment matter for legacy modernization?

Regulated industries operate under strict data governance frameworks that fundamentally restrict how information can be processed and stored. Financial institutions, insurance carriers, energy operators, and public sector agencies cannot simply route their core databases through public cloud infrastructure to access external artificial intelligence capabilities. Compliance regulations mandate that sensitive customer data, transaction records, and proprietary business logic remain within controlled organizational boundaries. This creates a significant barrier for modern AI development tools that rely on cloud-based processing. The most valuable data for modernization efforts is precisely the data that compliance rules keep isolated from external networks.

Companies that attempt to bridge this gap often encounter security audits that halt progress, legal reviews that delay deployment, and internal resistance from risk management teams. The architectural decision to run entirely on-premise, within virtual private clouds, or in fully air-gapped environments is therefore not merely a technical preference but a fundamental requirement for enterprise adoption. The founders of Kodesage recognized this constraint early in their development process. Gergely Szilagyi, who previously co-founded the encrypted storage company Tresorit, brought a design philosophy centered on data sovereignty to the project. The platform operates by deploying directly into the customer environment rather than requesting external access. Source code, database schemas, and configuration files never leave the organization’s control during the analysis and conversion phases.

This architecture aligns with the procurement requirements of large enterprises that demand complete visibility and control over their security posture. It also addresses a practical financial consideration by detaching operational costs from per-token cloud pricing models. Enterprises wary of unpredictable metered bills find value in a predictable infrastructure model that scales with their internal compute resources rather than external API calls. This economic alignment matters significantly for industries that must justify technology expenditures through rigorous return-on-investment calculations. The initial deployment of an on-premise platform requires capital allocation for hardware and installation, but the ongoing operational costs remain stable regardless of how much code is processed. This predictability simplifies the approval process for chief information officers and finance directors who must balance innovation budgets against operational stability.

How does automated discovery transform buried business logic?

Legacy codebases rarely organize their business rules in clean, modular files. Instead, the actual logic sits buried within stored procedures, database triggers, configuration files, and tightly coupled application layers. Extracting this information requires a systematic approach that can parse complex dependencies without disrupting active production systems. The platform performs what the company describes as automated deep discovery, analyzing the structural relationships between different components to reconstruct the underlying business logic. This process generates a living documentation layer that both human development teams and artificial intelligence agents can interpret. The documentation is not a static report but a dynamic reference that updates as the system evolves. From this foundation, the platform supports context-aware code conversion, which maps outdated constructs to modern equivalents while preserving functional behavior.

The market entry strategy focuses on a specific technical wedge that remains widely deployed across regulated sectors. Oracle Forms continues to power numerous enterprise applications despite being largely superseded by newer development frameworks. Mainstream tooling often provides inadequate support for migrating this particular stack, leaving organizations to manage the transition manually. Kodesage offers a dedicated modernization kit designed to accelerate the conversion process, such as migrating from Oracle Forms to Oracle APEX. The company claims these migration recipes can accelerate conversions by up to three times while reducing documentation effort by more than eighty percent. These figures represent internal company metrics rather than independently audited results, but they highlight the potential efficiency gains when automated analysis replaces manual reverse engineering. The platform also generates automated test suites to validate the converted code, ensuring that functional accuracy is maintained throughout the transition.

The Economic Shift Away from Tokenized Cloud Pricing

Enterprise software procurement operates on fundamentally different timelines and budget structures than consumer technology adoption. Large organizations plan their technology investments years in advance and require predictable cost structures that align with fiscal planning cycles. The prevailing model for cloud-based artificial intelligence tools relies on metered usage, where costs scale directly with the volume of tokens processed. This pricing structure creates uncertainty for enterprises managing massive codebases, as the financial outcome of a large-scale modernization project becomes difficult to forecast. On-premise deployment eliminates this variable by shifting the cost model toward infrastructure ownership and licensing. Organizations can allocate their existing compute resources to handle the analysis workload without worrying about fluctuating API fees.

This economic alignment matters significantly for industries that must justify technology expenditures through rigorous return-on-investment calculations. The initial deployment of an on-premise platform requires capital allocation for hardware and installation, but the ongoing operational costs remain stable regardless of how much code is processed. This predictability simplifies the approval process for chief information officers and finance directors who must balance innovation budgets against operational stability. It also reduces the friction associated with scaling modernization efforts across multiple departments or geographic regions. When the financial model does not penalize increased usage, organizations can pursue comprehensive architectural upgrades rather than piecemeal fixes. The shift away from tokenized pricing reflects a broader industry trend toward enterprise-grade artificial intelligence that prioritizes long-term planning over short-term consumption metrics. The competitive landscape for artificial intelligence-driven modernization is rapidly intensifying. Numerous technology providers are attempting to apply machine learning to enterprise software challenges, but success will depend on trust, compliance, and architectural alignment.

What lies beyond automated migration?

The immediate objective of the platform is to accelerate the conversion of outdated systems into modern architectures. The longer-term vision extends further into the operational lifecycle of enterprise applications. The founders describe an ambition to create self-healing enterprise applications where artificial intelligence agents continuously monitor system behavior, prepare fixes, generate tests, and implement updates. Human engineers would transition from writing code from scratch to reviewing, approving, and overseeing automated maintenance cycles. This represents a fundamental shift in how organizations manage their software portfolios. Instead of treating legacy systems as static liabilities, institutions would view them as dynamic environments that can adapt to changing requirements with minimal manual intervention. The company frames this direction of travel as a strategic goal rather than an immediate product offering, acknowledging the technical and regulatory hurdles that must be addressed before such autonomy becomes viable.

The recent seed funding will support a go-to-market expansion across the United States and Europe, alongside continued hiring in engineering and product development. The company enters a crowded market with a clear differentiator: a platform designed from the ground up for the constraints of regulated industries. The challenge ahead involves proving that automated discovery can reliably capture complex business logic while maintaining the rigorous accuracy standards required by financial and insurance sectors. The outcome of this effort will influence how enterprises approach the ongoing transition from legacy infrastructure to modern computational paradigms.

Conclusion

The modernization of core enterprise software represents a structural challenge that extends far beyond individual technology stacks. Institutions that manage critical financial and operational processes must balance the need for innovation with the imperative of stability. The deployment of artificial intelligence within secure, on-premise environments offers a pathway to address decades of accumulated technical debt without violating compliance requirements. As regulatory frameworks continue to evolve and computational capabilities advance, the tools available to engineering teams will shift from manual reconstruction to automated analysis. The success of this approach will depend on sustained investment, rigorous validation, and a clear understanding of the operational realities that govern large-scale software management. The industry is moving toward a model where legacy systems are not abandoned but systematically transformed, allowing organizations to preserve institutional knowledge while upgrading their technical foundations.