systemsEnglishOnline Dieses Seminar auf Deutsch →
This hands-on workshop brings product owners, business analysts, requirements engineers, architects and AI agents together at one table. Supported by AI agents, you learn to create and refine a specification that can serve as the basis for AI code generation and testing. Our approach, systems42, combines good practices from the proven req42 and arc42 approaches – open source and pragmatic.
Working with AI in several short iterations, you build a systems42 specification for a case study: from vision and goals, via user stories, to concrete, testable specifications. You review and extend this spec iteratively and incrementally.
To finish, you see live how an AI agent generates code and tests from your specification – traceable back to your requirements.
- Duration
- 2 days
- Participants
- 6–12 people
- Price
- €1,500 per person, €1,350 for alumni or early-bird bookings (see registration)
- Language
- German / English
- Who it is for
- Product owners & business analysts
- Software architects & tech leads
- Requirements engineers & developers
- Testers & QA engineers
- Teams that want to steer AI agents through good specifications
- Organisations that want to work in an agentic, AI-supported way for the long term
- Trainers
- The creators of systems42
Peter Hruschka
Gernot Starke
Peter Gerstbach - Format
- Public
- In-house
- Online
Topics
Day 1 – Developing the core spec with agentic engineering
- Introduction
- Why good specs matter more than they used to
- What is systems42?
- Goals, scope & stakeholders
- Capturing raw input with AI
- Detailing functional requirements
- Developing a glossary & quality scenarios
- Creating a story map & prioritising pilot stories
Day 2 – Spec-ready stories & live demo
- Specifying acceptance criteria for requirements
- Solution strategy, architecture & ADRs
- Inspect & Clarify: reviewing and sharpening the spec
- Live demo: the agent generates code & tests from your spec
- Transfer: the spec workflow in your own team
What you take away
- A systems42 specification in your repository
- Spec-ready requirements with acceptance criteria
- Prompt templates & rules for AI-supported specification
- Traceability: requirement › spec › code › test
- An agentic workflow with Capture, Inspect and Clarify loops
Prerequisites
Some experience with AI assistants (e.g. Claude, ChatGPT). No programming skills required. Bring your laptop and join in – or simply follow the results.