Strategy
Clarity before code
Use-case portfolio with value × risk. EU AI Act classification. Investment plan across 12–24 months. What you present to your board after four weeks.
Services
Six services. Three fields.
Advisory, AI architecture, on-premise and air-gapped — bookable individually or as a programme. EU-compliant, on-prem, no vendor lock-in.
How we structure mandates
Clarity. Build. Adoption.
Strategy without engineering stays PowerPoint. Engineering without adoption stays a demo. We deliver all three — or exactly the one you're missing.
Build
From notebook to production
Pipelines, eval suite, monitoring. Audit-ready documentation per Annex IV. Weekly demos on real production code.
Enable
Making it stick
90-minute briefings for the board. 2-day bootcamps for engineering. Fractional lead when you're between programme and hire.
Service catalogue
Six services, three fields.
From the honest entry point to sovereign delivery. Fees scoped per project — discovery PoC, fixed price, day rate or retainer.
Advisory
AI Consulting & Transformation
Use-case prioritization, roadmap, enablement and fractional AI lead. The honest entry point instead of burning money.
Mehr erfahrenAdvisory
AI Readiness & EU AI Act
Readiness check, risk tiering, Annex IV documentation, audit prep. The deadline is 2 August 2026 — we'll get it done.
Mehr erfahrenArchitecture
Enterprise AI Architecture
Data foundation + traceable decision layer: Explain · Forecast · Simulate · Recommend — with a source per number, on-prem.
Mehr erfahrenArchitecture
LLM Integration
LLMs and governed agents in your processes — RAG, research, memos. Auditable, EU-hosted — built into your systems.
Mehr erfahrenOperations
On-Premise LLM
Language models in your own infrastructure — on-prem or in your cloud. Full data sovereignty, no data outflow.
Mehr erfahrenOperations
Air-Gapped AI
Fully network-isolated AI for defence, critical infrastructure and classified work. No internet, no external APIs.
Mehr erfahrenHow we work with you
Four phases. You can exit after any one.
No pitch for an 18-month programme. Each phase stands alone with its own output and its own invoice.
- 01
2 weeks
Discover
Stakeholder interviews, data audit, use-case inventory. Ends with a written recommendation — which sometimes says: don't build.
- 02
4 weeks
Design
Architecture decisions, risk tiering, compliance mapping. Blueprint on paper before anyone writes code.
- 03
8–16 weeks
Build
ML engineering, pipelines, eval suite. Weekly demos on real code. No big-bang reveal at the end.
- 04
ongoing
Operate
Monitoring, incident response, model lifecycle. You take over — we stay on-call for six months.
Frequently asked questions
Before you ask.
How quickly can you start a programme?
Kick-off within 1–3 weeks of signing. If procurement takes longer, the Discover phase runs in parallel — we're used to regulated industries.You say 'no platform lock-in' — what does that mean in practice?
We choose per use case between proprietary (OpenAI, Anthropic), open-weight (Llama, Mistral) and vendor-specific platforms (Foundry, Databricks). If you already have a platform, we integrate there. If not, we won't push one on you.Can you deliver strategy only, without a build?
Yes. Strategy mandates are bookable independently (from 4 weeks, fee on request). We only build if the strategy justifies it — and we explicitly say no if the use case is better left unbuilt.What does a typical mandate cost?
Every mandate is individually scoped. Three models: fixed price per phase (for clearly bounded scopes), day rates by seniority level (junior to principal), or retainer (for Operate and Fractional Lead mandates). Fee on request.Which industries are your home ground?
Mid-market and Industry 4.0, education and public sector, insurance and finance. We don't take mandates where we can't understand the domain in the first Discover week.How do you differ from Palantir + Bain?
They bind you long-term with high entry ticket and Foundry licence. We start with smaller, clearly bounded mandates, are vendor-neutral, and include the Annex IV documentation for EU AI Act audits.
First conversation
30 minutes. Concrete output.
We listen, we sort through it together. At the end we tell you whether AI is the right tool — or which one fits better.