Contents
Good preparation decides whether an agentic coding workshop is productive or gets bogged down in login problems and debates about principles. Teams that settle goals, tool approvals, data protection questions, the repository and practice tasks in advance gain valuable time for real hands-on work. This article shows what your team should do before the workshop – including a checklist.
Why Preparation Matters So Much
In agentic coding training, participants work with coding agents that read code, run commands and propose changes. That requires working access, settled permissions and suitable practice material. If any of these is missing, the first hour is spent troubleshooting. Unresolved questions of principle are even more costly: if someone asks mid-workshop whether company code may be sent to a cloud service at all and nobody has an answer, the hands-on work comes to a halt.
Preparation has three layers: organisational (goals, participants, setting), legal and security (approvals, data protection) and technical (tools, repository, tasks).
Set Goals and Expectations
Before turning to technology, be clear about what the workshop should achieve. Useful questions:
- Should the team first get to know agentic coding, or introduce a concrete workflow?
- Is there a project that needs the new way of working right away?
- What outcomes do you expect – knowledge, team rules, first delivered tasks?
- Who decides on further use after the workshop?
A short survey among participants helps gauge prior knowledge: who already uses AI tools, who is sceptical, what are the expectations? The trainer can then tailor content and exercises. To assess how ready your team is overall, try an AI readiness check for development teams.
Clarify Approvals, Data Protection and Security
This is the most common stumbling block. Before the workshop, settle the following with IT security, your data protection officer and, where relevant, the works council:
- Which tools are approved? Cloud-based agents, IDE extensions or locally hosted models?
- Which code may be used? The full repository, selected modules or practice projects only?
- Which contracts are needed? For example a data processing agreement with the tool provider if personal data could be involved.
- How are secrets protected? No production credentials in the practice environment, API keys with restricted rights.
- Which commands may an agent run? Decide on approval mode and sandboxing beforehand.
For background on whether and under what conditions source code may be processed by cloud tools, see AI coding tools and data protection. This article is not legal advice; involve your data protection officer if in doubt.
Technical Preparation: Tools, Access and Repository
A smooth start requires every participant to be able to begin immediately on the day. A short technical check a few days ahead has proven useful:
- Licences or API access are set up and tested.
- The tool starts on every machine, including behind the corporate proxy or VPN.
- The practice repository can be cloned, built and tested – locally or in a prepared container environment.
- The test suite passes and finishes in reasonable time. Agents need fast feedback.
- For remote workshops: video conferencing, screen sharing and breakout rooms work.
Also check whether the repository already contains project documentation, such as a README with build instructions or architecture notes. This information is the basis for the context files used to steer agents. How to write them is covered in writing project rules and context files for coding agents.
Choosing Practice Tasks
The best workshops use tasks the team knows from daily work. Together with the trainer, select a few tickets that meet these criteria:
| Criterion | Why it matters |
|---|---|
| Clearly bounded | The agent needs an unambiguous goal and a verifiable definition of done. |
| Solvable quickly | The workshop should allow several iterations, not one endless attempt. |
| Testable | Existing or quickly added tests make results verifiable. |
| Non-critical | No tasks requiring access to sensitive data or production systems. |
| Varied difficulty | A mix of simple and harder tasks reveals strengths and limits. |
Good candidates include small features, bug fixes with a reproducible error, missing tests for an existing module or a bounded refactoring. Why working on your own code is so effective is explained in custom AI workshops on your own codebase.
Checklist: The Final Days Before the Workshop
- Goals and expected outcomes are written down and agreed with the trainer.
- Participant list is final, prior knowledge has been surveyed.
- Tool approval from IT security and data protection is in place.
- Access is set up and tested for everyone.
- Practice repository is prepared; build and tests run.
- Three to six suitable tasks are selected and described.
- Room or remote setup is organised; participants' calendars are blocked.
- A follow-up date has already been scheduled.
The last item looks like a detail but is decisive: when the follow-up is already in the calendar, what was learned is more likely to be applied. Do not underestimate calendars either. Participants who handle support tickets on the side benefit far less.
How the Trainer Supports Preparation
An experienced trainer takes on part of the preparation, for example through a kick-off call, reviewing the repository or suggesting suitable exercises. At app-entwickeln-lassen.de we place AI developers from real-world practice as trainers; content, format and scope are agreed with you individually. Workshops start from €799 net, and you receive a concrete quote on request by e-mail.
Planning an app? Request a free fixed-price quote in 2 minutesConclusion
Careful workshop preparation saves time during training and significantly increases the learning effect. Settle goals, approvals and data protection in advance, test the technical environment and select real, well-bounded tasks. Your team can then focus on what matters in the workshop: actually practising agentic coding.
Frequently Asked Questions
How much lead time does workshop preparation need?
That mainly depends on internal approval processes. Allow enough time for alignment with IT security and data protection, as this often takes longer than the technical setup.
Can we run the workshop without access to our own code?
Yes, with prepared practice projects. Transfer to daily work usually works better with your own code, however.
Do participants need to learn anything beforehand?
Not necessarily. It helps if everyone has launched the chosen tool once and tried a simple task, so technical problems surface early.
Who should be involved in preparation?
At least the tech lead, IT security and, where relevant, the data protection officer. In larger organisations it also makes sense to inform the works council early.