AI strategy, GenAI implementation, AI agents, MLOps and EU AI Act compliance — at the level of top-tier international consultancies, integrated with tax, legal and holding expertise under one roof. For mandates where AI is meant to become a production system, not just a "pilot".
We don't build AI mandates as pilot projects that never leave the lab — we build production systems with clear value creation, regulatory resilience and a scaling roadmap. Senior-partner-led, vendor-independent, backed by TABAK's tax, legal and holding expertise.
Where does AI pay off — and where doesn't it? A structured readiness assessment across data, process and technology maturity. Use-case prioritisation with business case and risk matrix.
Production-grade integration of GPT, Claude, Llama and domain-specific models — RAG architectures, prompt engineering, function calling, tool use, hallucination control. End-to-end, from POC to production.
Autonomous workflow agents, multi-agent systems, long-horizon tasks. We don't build toy chatbots — we build agents that take on real value creation: controlled, auditable, with escalation paths.
Classic RPA combined with AI models for hybrid automation. Document intelligence, OCR with genuine comprehension, decision-making in unstructured data.
ML and deep learning models for specialised use cases. Fine-tuning of large language models on domain-specific data, vision models, forecasting and anomaly detection systems.
Applicability analysis under the EU AI Act, high-risk system classification, conformity documentation, risk management frameworks. With verifiable audit trails — before the regulators come knocking.
Data architecture as the foundation: data lakes, feature stores, ML pipelines, model monitoring, A/B test setups. Without a clean data infrastructure, every AI initiative fails.
Executive briefings, technical training for development teams, AI compliance workshops for business units. Including in-house sessions at C-level.
Combine advisory modules and calculate your fee in real time. Fixed fees for initial mandates, retainers for ongoing support — at the level demanding AI mandates require.
Your AI mandate briefing has been received. We'll be in touch within one business day for a confidential scoping call.
We don't build AI demos that gather dust after the board presentation. Every POC comes with a scaling path, MLOps infrastructure and a compliance roadmap. That sets us apart from many pure-play AI boutiques — and makes AI at TABAK an investment-grade business area.
Structured identification of the right AI use cases along value creation and data availability. With a real business case — not "because it's trendy right now".
Even at POC stage, we build with the libraries, structures and security standards that will apply in production. No throw-away code.
EU AI Act, GDPR, industry-specific requirements — factored in from the scoping phase. We don't build systems that later have to be shut down as "non-certifiable".
Model monitoring, drift detection, re-training pipelines. Without these structures, production models degrade within 6–12 months. We always build them in.
Managing directors and boards who need to prepare AI initiatives — for competitiveness, efficiency or the investor story. We deliver strategy and production-grade POCs that can be explained internally and externally.
Insurers, banks, healthcare, critical infrastructure — clients whose AI must operate under the AI Act's high-risk classification. We make them audit-proof.
AI due diligence for investments, valuation of AI-driven business models, post-acquisition AI integration. With the discretion these mandates demand.
Independently operating group subsidiaries that need their own AI roadmap — aligned with group compliance, but technologically self-reliant.
Both — and that is precisely one of our strengths. For many use cases, frontier models (GPT-5, Claude, Gemini) are the right choice. For regulated data, offline requirements or specialist domains, we develop custom models or fine-tunings. The use case makes the decision — not a vendor bias.
Readiness assessment: 4–6 weeks. GenAI pilot with production path: 3–4 months. Custom model development: 4–9 months. EU AI Act compliance for high-risk systems: 3–6 months. We commit to a binding timeline during the scoping phase.
We don't just build the use case — we think through the tax, legal and holding context alongside it. Implementing AI in a regulated industry means AI Act compliance, GDPR records of processing activities, and potentially IP structuring for the models developed. We deliver all of this under one roof — no agency can do that.
Initial mandates start at approx. €18,000 (readiness assessment). Implementation mandates from €45,000. We deliberately decline small engagements — the depth AI demands today cannot be delivered credibly below a certain volume.
Both. Cloud (Azure OpenAI, AWS Bedrock, Anthropic via API) for most commercial use cases. On-premise or sovereign-cloud setups (Llama, Mistral models, local inference) for regulated data, critical infrastructure and clients with data protection concerns about external providers.
Initial consultation with no obligation. NDA in advance on request. We listen before we propose.
