Most AI-built software breaks before it ships

Teams adopting agentic AI are hitting a wall: once a project reaches critical mass, regression rates climb past 50%. Every new sprint breaks something that worked yesterday. The code gets written fast — and then collapses under its own weight, buried in refactoring.

The problem isn't the AI. It's the absence of DevOps discipline built for agentic development. Structured documentation, context management, agent personas, and a real pipeline aren't overhead — they're the difference between software that works once and software that works forever.

Over two days, you'll build that pipeline yourself using Claude Desktop Project Agent, Claude Code Developer, and Cursor Peer Review Agent — then use it to ship a working Industry 4.0 tool to GitHub. By the end, you won't just have a tool. You'll have a repeatable way to build software that holds up.

Who this course is for

• Manufacturing end-users — engineers, process leads, and OT professionals who want to build their own tools without a software development background
• System integrators building custom solutions on top of purchased platforms (Ignition, MQTT brokers, MES, ERP connectors, DataOps platforms)
• IT/OT professionals managing Industry 4.0 infrastructure who need to extend it with custom tooling

Assumes familiarity with Industry 4.0 and IIoT concepts. No professional software development experience required.

Schedule

Workshop Dates: July 21–22, 2026 | 9:00 AM – 1:00 PM CDT (each day)

Day 1: Theory & DevOps Pipeline Build — Tuesday, July 21

• The Post-Agentic World: Why end-users now build their own software instead of waiting on vendors
• Documentation, Context & Agent Personas: The foundation that keeps AI-built software stable at scale
• Hands-On — Building the DevOps Pipeline: Configure Claude Desktop, Claude Code, and Cursor into a working agentic workflow
• Product Selection: The cohort defines what we build on Day 2

End of Day 1: A fully configured agentic DevOps pipeline and a cohort-selected product spec, ready to build.

Day 2: Hands-On Development — Wednesday, July 22

• Product Scaffolding: Repo setup, structure, and initial context load
• Core Development: Build the product live alongside the instructor
• Integration, Testing & Peer Review: Run a full Cursor peer-review pass to catch regressions early
• Final Polish & Publish: Ship to a private GitHub repo, then debrief

End of Day 2: A working Industry 4.0 software tool published to a private GitHub repository, accessible only to registered students.

Note: Session recordings will be provided to all registrants and posted to iiot.university.

What you'll learn

• Set up a complete agentic DevOps pipeline (Claude Desktop + Claude Code + Cursor)

• Author and manage project documentation that serves as live context for AI agents

• Define and configure agent personas for distinct roles in your development workflow

• Design DevOps workflows that minimize regression and refactoring

• Scope and build a real Industry 4.0 software tool with a cohort, in a single day

• Ship a production-oriented tool to GitHub using the full pipeline

• Apply best practices for context management, version control, and peer review

Requirements

• Active Claude account (Pro or Team) with Claude Desktop installed
• Cursor installed and authenticated with Claude
• Claude Code enabled in their environment
• GitHub account — username submitted to the team before the workshop
• A laptop capable of running local development tools
• Working familiarity with Industry 4.0 / IIoT concepts (UNS, MQTT, edge/cloud architecture, basic data flows)