Maintenance, defined.

Technical debt is the accumulated cost of outdated components, shortcuts and deferred maintenance in a software system. Today it spans the whole engineering stack: runtimes reaching end of life, infrastructure aging out of support, dependencies drifting out of compatibility, configuration drift and services without clear owners. Left unmanaged, it surfaces as outages, failed audits and rising cloud costs.

Technology lifecycle management is the continuous practice of tracking every technology in your stack against its vendor’s lifecycle (general availability, end of standard support, end of life) and upgrading before support ends. It covers operating systems, databases, runtimes, frameworks, managed cloud services, certificates and AI models.

End of life (EOL) is the point after which a vendor no longer maintains a software version: no security patches, bug fixes or support. Running EOL software increases security risk, can fail compliance controls such as PCI DSS and SOC 2, and can trigger extended-support fees.

End of support (EOS), often called end of standard support, is when a vendor stops providing standard support for a version. Many vendors then offer paid extended support for a limited period before the version reaches end of life.

Extended support is a paid option, offered by many cloud providers and software vendors, to keep receiving security fixes for a version after its standard support ends. Because costs often scale with usage, delayed upgrades can add significant, unplanned spend.

A software factory is an organization’s end-to-end system for producing software: plan, design, build, review, test and ship. In an AI software factory, platforms and coding agents automate more of that build line. Everything it ships still has to be maintained after release.

A maintenance layer is the part of the software factory responsible for keeping everything already running current, secure and compliant. It works as a continuous loop: identify what changed, understand the impact, prioritize, act, verify the fix and govern the process.

MaintenanceOS is Draftt’s platform: the maintenance layer that connects to your existing tools, identifies what changed, prioritizes what needs attention and coordinates your teams and AI agents to fix it under your policies and approvals.

Autonomous remediation uses automation and AI agents to fix issues such as vulnerabilities, outdated dependencies and misconfigurations end to end, from detection to a verified change. Governance decides which changes may run on their own and which need human approval.

Configuration drift is when the actual configuration of systems diverges from the intended or approved baseline over time, for example backups disabled, encryption settings changed or versions differing between environments.

Continuous compliance means keeping systems aligned with frameworks such as PCI DSS, SOC 2, ISO 27001 and HIPAA at all times, instead of preparing for each audit. Much of the work is engineering: upgrading unsupported software, patching vulnerabilities and fixing configuration gaps, with evidence retained.