EU AI Act –
From Paragraph to Pull Request

A company prepares for the EU AI Act by systematically creating clarity before implementation begins.

Concretely, this means: First, identify which AI systems exist or are planned. These are classified to understand which ones fall under the EU AI Act and where heightened requirements arise. Then, derive which technical minimum requirements are relevant and which obligations apply now — and which do not yet.

Only then are requirements translated into concrete tasks, prioritized, and fed into the backlog. This turns regulation from an abstract risk into actionable work that blocks neither development nor operations.

For AI teams, the EU AI Act primarily means additional technical obligations in day-to-day development and operations — not new theory.

Specifically, teams must understand which of their systems are affected, how they are classified, and which minimum requirements must be met technically. This includes requirements for data provenance and documentation, model and system documentation, logging and monitoring in production, evidence for decisions, and clearly defined forms of human oversight.

The key difference from before: Many things already done technically must now be done deliberately, traceably, and verifiably. For AI teams, the EU AI Act becomes a question of structure, prioritization, and solid engineering — not a purely legal matter.

The effort depends heavily on the specific AI use case. Not every AI system falls under the regulation, and not every system faces the same obligations.

The higher a system's risk class, the more extensive the requirements. AI systems in regulated areas or higher risk classes require significantly more evidence, documentation, and organizational measures than simple or supporting applications. For other systems, the additional effort is correspondingly lower.

At the same time: The EU AI Act's requirements aim less at new technology than at structure, traceability, and proper classification. The effort arises primarily where it's unclear which obligations concretely apply and how they should be sensibly implemented.

What matters is not a blanket assessment, but precise classification per use case: What is affected, which minimum requirements apply — and where there is no need for action. This is exactly where the AI Act Crash Course comes in.

The AI Act Crash Course is designed for companies that develop, operate, or productively deploy AI systems and now need to clarify what the EU AI Act concretely means for their technology and operations.

The format is particularly suitable for AI, data, and engineering teams, tech leads, product owners, CTO-adjacent roles, and interfaces to legal and compliance. In other words: for everyone who bears responsibility for AI systems and needs to know what must be implemented technically and what can wait.

The crash course is not intended as a pure management or theory course. It's for teams that build, deploy, and operate AI — and need a clear, actionable classification of AI regulation for that.

The right time for an AI Act Crash Course is whenever AI already exists or is concretely planned — and there is uncertainty about what the EU AI Act means for it.

Typical triggers include:

• AI systems are in production or close to go-live
• New AI use cases are being developed or scaled
• Questions from legal, compliance, or management are increasing
• Decisions are being postponed or blocked out of caution
• It's unclear which minimum requirements need to be implemented now

In short: Before regulation slows down development and operations — but after it's clear that AI is a real, active topic in the organization.

The AI Act crash course is not classic training, but a working format for AI teams. Normal training provides knowledge about the EU AI Act. The crash course translates regulation into concrete relevance for your AI systems.

The difference in practice:

• Not a foil battle, but a classification of real AI applications
• Not a theory, but a derivation of minimum technical requirements
• No general statements, but a clear distinction between must-haves, should-haves and nice-to-haves
• Not a certificate without context, but knowledge that fits directly into the backlog

In short: Training explains the EU AI Act. The AI Act crash course makes it manageable for the development and operation of AI systems.

An AI team takes away clarity and the ability to act from the AI Act Crash Course. Concretely, this means:

• A clear classification of what the EU AI Act means for their existing and planned AI systems
• Clarity on which requirements are relevant now and which are not
• A shared understanding of what counts as technically "compliant"
• A concrete list of technical minimum requirements for data, models, monitoring, logs, and documentation
• Actionable to-dos as a basis for tickets, backlog, and prioritization
• Confidence in communicating with legal, compliance, and management

In short: After the crash course, the team knows not only what the EU AI Act requires, but what concretely needs to be done — and can get back to working on the product.

The AI Act Crash Course takes one day. The workshop is intentionally kept compact and focuses on what matters: clear classification, concrete requirements, and directly actionable technical to-dos. No multi-day training, no theory blocks — just a concentrated working day with real output for the AI team.

The people who bear professional, technical, or organizational responsibility for AI systems and who prepare or implement decisions should attend.

Typically, these are:

• Heads of AI, data, or ML teams
• Tech leads, engineering leads, and product owners with AI focus
• Roles responsible for operations, monitoring, or quality assurance
• Interfaces to legal, compliance, or data privacy

What matters is not legal background, but proximity to implementation. The greatest value comes when technology and the relevant governance roles are in the room together.

No extensive preparation is needed on your side, but a targeted exchange beforehand is important. This includes gathering the relevant AI use cases, models, or systems currently in use or concretely planned.

Additionally, we conduct preparatory conversations with key stakeholders from the AI team, engineering, legal, or compliance. In these conversations, we clarify the current situation, expectations, and specific questions. This ensures the AI Act Crash Course is precisely tailored to your organization and focuses entirely on the truly relevant topics during the workshop.

The partnership with Taylor Wessing means that legal classification and technical implementation go hand in hand from the start in the AI Act Crash Course. Regulatory questions are not reviewed after the fact, but clarified directly in the context of AI systems, architecture, and operations.

In the crash course, Fritz-Ulli Pieper, Salary Partner at Taylor Wessing and expert in AI, IT, and data privacy law, brings the law firm's perspective. He provides the legal classification of EU AI Act requirements and, together with PLAN D, translates them into concrete, technically manageable guardrails.

For you, this means: Technology, regulation, and legal responsibility are not discussed separately. AI teams receive a legal classification that is directly translated into technical to-dos, backlogs, and operational decisions. Legal knows what matters. Engineering knows what to implement.

PLAN D combines technical understanding, proximity to implementation, and regulatory classification in an integrated approach. The benchmark is not interpretation, but implementability.

What sets us apart:

• Technology from practice: We speak the language of developers, data scientists, and AI teams because we come from technical implementation ourselves. Architecture, data, models, and operations are our daily work.

• Legally considered: Through the partnership with Taylor Wessing, regulatory questions are classified early and flow directly into strategic and operational decisions.

• Experience from real projects: We have been working on production AI systems for years — from development through operations to scaling, in mid-market companies as well as enterprises.

• A team, not a toolkit: You work with clearly named contacts who take responsibility — in the workshop and beyond.