008 Alexander Zeitler - Running a Manufacturing Company on Omarchy & AI Agents
Summary
Alexander Zeitler, a software developer and manufacturing company owner, shares his journey from legacy Windows systems to Omarchy and AI-driven development. He details using AI agents to extract data from proprietary ERP software and build custom Rails applications for engineering documentation and decision tracking. The episode explores his unique workflow using GitHub issues and event sourcing to manage agent interactions, emphasizing the importance of domain knowledge and structured reasoning for future software engineers.
Topics discussed
Introduction and Alexander's early programming history
Transition to Linux and Omachi, and early business software
Recollections of the pre-internet era and LAN parties
The struggle with legacy ERP software and data access
Using AI to extract data and the principle of data ownership
The 'golden cage' of proprietary vendors like Microsoft and SolidWorks
Rethinking CAD with open tools like FreeCAD and OpenSCAD
Building internal software for decision recording and collaboration
Documenting engineering decisions and managing product changes
Automating compliance documentation with AsciiDoc and components
Coding entirely with AI agents and learning Rails
Deep dive into Event Sourcing architecture and F1 example
Challenges of agent autonomy and the need for structured workflows
Using GitHub issues and labels to manage agent interactions
Security concerns, context poisoning, and agent code review
Essential skills for software engineers in the AI era
Domain-Driven Design, Event Modeling, and GWT syntax
The immutability of decisions and historical state persistence
Applying Event Sourcing to the GitHub development process
Benefits of parallel agents and cross-issue reasoning
Closing remarks and future follow-up
Listen ad-free on Castria