Responsible AI and the EU AI Act for the Mittelstand.
EU AI Act compliance, Responsible AI and GDPR conformity: three promises, one solution. With on-premise AI, you keep full control over your data.
5 minutes, 14 questions, free.
At ProzessX, Responsible AI means introducing AI in the Mittelstand in a way that complies with the EU AI Act and the GDPR. We classify your AI systems by risk category, produce the required technical documentation, and train your team on AI literacy under Art. 4, in force since February 2, 2025. With on-premise AI, your data stays inside your own company.
Caught in the middle as an SME.
GDPR violations loom
US cloud providers are subject to the CLOUD Act: US authorities can access your data. A clear conflict for Responsible AI. The GDPR provides for fines of up to 20 million euros or 4% of global annual turnover (Art. 83 GDPR).
EU AI Act deadlines are running
Since February 2, 2025, the obligation to ensure AI literacy (Art. 4) has applied: companies must take measures so their staff use AI competently. Since August 2, 2026, the transparency obligations apply. The high-risk requirements take effect, following the Digital Omnibus, from December 2, 2027.
Technological dependency
Price increases, API changes, service outages: relying on Big Tech means giving up control. Avoid vendor lock-in.
Data protection that doesn't slow you down.
- Your data stays in the EU or on your own hardware (on-premise).
- No AI output leaves your business without sign-off from a person.
- We classify your AI systems under the EU AI Act and train your team.
Three pillars of Responsible AI.
ProzessX Consulting
Strategic consulting for your AI deployment.
- EU AI Act readiness assessment
- Responsible AI strategy
- GDPR-compliant implementation
- Change management
AI Academy
Training for employees and leadership.
- AI literacy training (Art. 4)
- "AI license" for employees
- Prompt engineering workshops
- Executive briefings
On-premise infrastructure
Physical infrastructure inside your own company.
- Neural Core Server
- Tier model, API to full-local
- Robotics integration possible
- Support from Nagold
What you need to know now.
Unacceptable AI practices
Social scoring, manipulative AI, real-time biometric identification in public spaces (with narrow exceptions), emotion recognition in the workplace. Prohibited since February 2, 2025.
Strictly regulated applications
Critical infrastructure, recruitment and HR management, creditworthiness assessment, education (Annex III): obligations from December 2, 2027. AI in regulated products such as medical devices (Annex I): from August 2, 2028.
Transparency obligations
Chatbots, AI-generated content, deepfakes. People must be able to recognize that they're interacting with AI or seeing AI-generated content (Art. 50). In force since August 2, 2026.
Largely unregulated
Spam filters, video games, industrial automation (without safety components). No specific requirements.
What ProzessX does for your compliance.
Risk classification
Classifying your AI systems by EU AI Act category.
Documentation
Producing all required technical documentation.
Risk management
Setting up an AI risk management system.
Training
AI literacy training under Art. 4 of the EU AI Act.
The deadlines are already running.
The EU AI Act is the world's first comprehensive legal framework for Responsible AI. Using a risk-based approach, it categorizes AI systems and defines clear obligations.
With the Digital Omnibus (Regulation (EU) 2026/1744, in force since July 27, 2026), the EU has pushed back the deadlines for high-risk AI. The AI literacy obligation and the prohibitions still apply, and the transparency obligations have applied since August 2026. As of October 29, 2026; not legal advice.
- February 2, 2025: prohibitions and AI literacy (Art. 4, Art. 5)
- August 2, 2025: obligations for providers of general-purpose AI models
- August 2, 2026: transparency obligations (Art. 50), general start of application
- December 2, 2027: high-risk AI under Annex III
- August 2, 2028: high-risk AI in products under Annex I
How much stays in-house?
Four levels from API Hybrid to Full On-Premise. API Hybrid: cloud APIs with local orchestration. Edge Processing: local inference, cloud training. Hybrid Sovereign: local models, optional cloud. Full On-Premise: fully local sovereignty. From level to level, more stays in-house.
API Hybrid
Cloud APIs with local orchestration.
Edge Processing
Local inference, cloud training.
Hybrid Sovereign
Local models, optional cloud.
Full On-Premise
Fully local sovereignty.
From API hybrid to full-local.
API hybrid
Cloud APIs with local orchestration.
- Fast start, low investment
- Local data pre-processing
- Migration to higher tiers possible
Edge processing
Local inference, cloud training.
- Sensitive data stays local
- Reduced latency
- Deployable in a GDPR-compliant way
Hybrid sovereign
Local models, optional cloud.
- Your own open-source models
- Full data control
- Cloud fallback possible
Full on-premise
Complete local sovereignty.
- No internet required
- Maximum compliance
- Data stays in-house
Choose the sovereignty tier for your Responsible AI infrastructure. We estimate the effort after the initial consultation.
Responsible AI for your industry.
Healthcare
GDPR, patient data and high-risk AI: maximum compliance requirements.
AI for medical practices →Automotive
Process supply chain AI and production data in an EU-compliant way.
AI for car dealerships →Everything you need to know about Responsible AI, the EU AI Act, and on-premise.
What exactly does "on-premise" mean for AI?
On-premise means the AI hardware sits inside your own company, not in a cloud. Your data never leaves your building, and no third party has access. That gives you real data sovereignty and makes you independent of the US CLOUD Act.
What funding is available for AI projects?
There are federal and state funding programs for consulting, investment and training. You'll find a vetted overview on our funding programs page. We help you find the right programs.
How long does implementation take?
How long it takes from the first consultation to go-live depends on the scope. We estimate it after the initial consultation. The path includes: needs analysis, hardware procurement and installation, software configuration, integration with existing systems, staff training, and go-live support.
Can I start with a cloud solution and switch later?
Yes, our tier model is designed exactly for that. You can start at tier 1 (API hybrid) and migrate step by step to tier 4 (full on-premise). Our solutions are built so an upgrade is possible without data loss.
What is the AI literacy obligation under Art. 4?
Article 4 requires providers and deployers of AI systems to take measures that promote the AI literacy of their staff. Since the Digital Omnibus, it's been clarified that no specific competence level for individual people has to be guaranteed. The obligation has applied since February 2, 2025. Our AI Academy offers suitable training.
How high are the penalties for violating the EU AI Act?
For prohibited practices, up to 35 million euros or 7% of global annual turnover; for violations of other obligations, up to 15 million euros or 3%; for false information to authorities, up to 7.5 million euros or 1% (Art. 99). For SMEs, the lower amount applies in each case.
Two paths to get started.
EU AI Act readiness check
Get a result in just 15 minutes.
- Which of your AI systems are affected
- Which risk category you fall into
- Which deadlines apply to you
- Which funding programs might fit
Consulting call
Or call us directly: +49 7452 6051953.
- 30 minutes, no obligation
- Individual analysis of your situation
- A concrete roadmap for your AI compliance
- Direct access to our AI strategists
What are you waiting for?
Let's find out where the biggest untapped potential in your company lies.

