Hypercubic AI shipped Hopper, an agentic development environment designed to modernize mainframe programming through AI assistance.

The desktop application combines a traditional TN3270 terminal with AI agents that can navigate IBM z/OS systems, write Job Control Language (JCL), debug failed jobs, and query VSAM databases using natural language prompts.

Hopper's AI agent understands mainframe-specific protocols like ISPF panel navigation and can decode complex system outputs including JESMSGLG, JESYSMSG, and SYSUDUMP files into structured diagnostics. The tool aims to reduce the hours typically spent troubleshooting through SDSF job logs.

Targeting legacy system modernization

The platform addresses a critical skills gap in mainframe development, where aging COBOL programmers are retiring faster than new developers are learning the technology. Hopper allows developers to interact with mainframe systems through modern interfaces while maintaining compatibility with existing z/OS infrastructure.

Key features include automated JCL generation, CICS deployment workflows, and the ability to compile, test, and deploy applications through single prompts. The agent pauses for user approval before making system changes.

Hypercubic AI's team includes former engineers from Cognition, Apple, Windsurf, Deloitte, ADP, and Volkswagen. The company offers free mainframe access through a partnership with Maintec for developers who want to test Hopper without existing z/OS credentials.

The software launches with a free "Hobby" tier supporting macOS, Windows, and Linux, plus an enterprise version with SAML SSO, admin controls, and on-premises deployment options. Pricing for the enterprise tier follows a custom model.

Hopper represents a broader trend of AI tools targeting legacy enterprise systems, similar to how Cursor and other AI coding assistants have modernized general software development workflows.

The company plans to expand Hopper's capabilities based on user feedback from the initial release.