(Software developers, software architects, team leads, product owners)
Requirements, design, build, test, deploy, maintain – that is how we built software for decades. That world no longer exists. In this 2-day workshop you build complete applications in dialogue with AI agents such as Claude Code and Codex – with the IDE closed and without reading a single line of code. What used to take days now takes minutes to hours. You experience this hands-on in your own project, not on slides.
We clear out the habits development teams are clinging to right now:
Up-front specifications are waterfall in disguise – we clarify intent and iterate.
arc42 and architecture documents are outdated the day they are finished – the Git history is the event stream of your decisions.
Clean Code was made for humans who read code – the AI structures the code itself, you secure responsibility through testability.
Human code reviews no longer scale – AI reviews AI.
What remains is what matters: the right decision between simple logic and a distributed system, clear responsibility through tests and APIs – and knowing where professionals will remain indispensable.
Learn:
Build applications in dialogue with AI agents – without an IDE and without reading code
Clarify requirements through intent instead of lengthy specifications
Use the Git history as living documentation and derive architecture decisions with a prompt
Take responsibility through testability, API first and AI reviews
Recognize when you need professionals: distributed systems, scaling, regulation
Iterate instead of plan: build, verify, throw away, rebuild – in minutes instead of weeks
Understand why software is becoming a commodity – and what role developers will play
Build applications in dialogue with AI agents – without an IDE and without reading code
Clarify requirements through intent instead of lengthy specifications
Use the Git history as living documentation and derive architecture decisions with a prompt
Take responsibility through testability, API first and AI reviews
Recognize when you need professionals: distributed systems, scaling, regulation
Iterate instead of plan: build, verify, throw away, rebuild – in minutes instead of weeks
Understand why software is becoming a commodity – and what role developers will play
References
Training
Content
Show details
Compact
1 Day
1d
Standard
2 Days
2d
The old lifecycle under scrutiny
Requirements, design, build, test, deploy, maintain: how we used to build software
Estimation round: how long does your team need for a time tracking system?
What AI changes in every single step
✓
✓ ✓ ✓
✓
✓ ✓ ✓
Requirements: intent instead of specification
Talk to the AI about goals, not forms
Role play: the trainer is the customer, the requirements come as a PDF
Let the AI analyze existing software: “How can this be done better?”
✓
✓ ✓ ✓
✓
✓ ✓ ✓
Build without an IDE
Claude Code, Codex & co.: software is created in dialogue
AGENTS.md as the only rulebook for the AI
Why we no longer need to read the code
✓
✓ ✓ ✓
✓
✓ ✓ ✓
Git history instead of architecture docs
Commits as an event stream of decisions
A prompt summary in every commit
Derive architecture decision records from the history with a prompt
Why arc42 and Clean Code have had their day
✓
✓ ✓ ✓ ✓
✓
✓ ✓ ✓ ✓
Practice 1: time tracking in hours instead of days
Vibe coding as a team, from requirements to a running application
The reveal: estimate versus reality
The new lifecycle: intent → iteration → verification → release
✓
✓ ✓ ✓
✓
✓ ✓ ✓
Design: local or distributed?
The one decision that determines the basic architecture
Simple logic: anyone can vibe code it
Distribution, scaling, enterprise: where professionals are still needed
X
X X X
✓
✓ ✓ ✓
Responsibility without reading code
Structure follows testability
API first: the UI is volatile, the API is the test bench
Have critical parts isolated so they stay testable and changeable
AI reviews AI: why human code reviews no longer scale
✓
✓ ✓ ✓ ✓
✓
✓ ✓ ✓ ✓
Practice 2: conference management
More roles, more data, more interfaces
Evolve, throw away, rebuild: maintenance is just the loop
Deployment by prompt, and where Docker, CI/CD and DevOps still matter
X
X X X
✓
✓ ✓ ✓
Spec-driven development in context
GitHub Spec Kit & co.: what is behind it
Why up-front specification is often waterfall in disguise
When a plan as an artifact still helps
X
X X X
✓
✓ ✓ ✓
Outlook: software becomes a commodity
Why classic software as a service is under pressure
Where classic development remains: regulation, physical products, high complexity
New roles: developers as architects, reviewers and vibe coding coaches
✓
✓ ✓ ✓
✓
✓ ✓ ✓
Conclusion & Certificate
✓ ✓
per participant, plus VAT and travel expenses as incurred
A training lights the spark – everyday work decides whether it keeps burning. With our vibe coding coaching we support your team in its first real projects: as much help as necessary, as little as possible.
Content adopted to your needs - for your team to profit in the best
possible way.
English and German - currently we offer this field-tested training in
whole Europe.
Experienced speakers known from various conferences (DWX, Magdeburger
DevDays, IT-Tage, ...) having the technical and dedactical skills to teach the exactly the knowledge
your scenario needs.
Inhouse or remote - all trainings.
A theory and a practice part is included in all trainings.
Complete training records will be handed over.
Support and consultancy will be provided even after the course.
A training certificate will be handed over to all participants.
My experience with the Clean Code training (2 days) was great. I gained a lot of knowledge about
clean code, refactoring, unit tests and various software design techniques. The trainer conveyed
the information in a very engaging way. There was always an opportunity to ask questions. I can
definitely recommend the Clean Code training for junior developers.
Henning
We have now completed our third training with Stefan Dirschnabel and it was once again a great
benefit for our team. Preparation top-notch, very structured process and excellent expertise. All
questions were answered and we were also offered help going forward. Thank you Stefan for the
training. We will definitely come back!
Johannes
Would do it again anytime! The Clean Code training was very well structured, had an enormous
amount of content, and was also wonderfully varied and entertaining. You can tell that Stefan
Dirschnabel brings not only a great deal of knowledge and experience, but also great enthusiasm
for software development!
You ask, we answer
Do participants need to be able to program?
For this workshop, yes: it is aimed at development teams and shows how their work is changing fundamentally. For domain experts without programming experience we offer the training “Vibe Coding for Domain Experts”.
Which AI tools do we use?
Mainly AI agents such as Claude Code and OpenAI Codex. The principles apply regardless of the tool; on request we work with the tools approved in your company.
Why does the training no longer cover prompt engineering and RAG?
Because today’s models and agents no longer need yesterday’s tricks. What matters is not the perfect prompt but a new development process: clarify intent, iterate quickly, secure responsibility through tests and APIs.
Can we bring our own projects?
Yes. In company trainings we like to work on your existing software: the AI analyzes it and we develop improvements together. The prerequisite is that the AI tools used are approved in your company.
Usage of Cookies
To optimize usage of our website and to provide you the best possible service and user experience, we use cookies.
By using this website you agree, that we set cookies on your device and are able to track usage of the website.
First-hand consultation
