Answer
What happens after I send the form?
We reply within one business day: times for a fit call, or availability for an assessment. You talk to the engineer who does the work. We confirm your stack and the blocker, agree scope, the access list and anything we should not see, and fix the price before kickoff. Sending the form does not commit you to anything, and if no fixed offer fits, we say so.
Answer
What does the assessment need from my team?
Usually a named technical owner and short interviews with engineering, platform, security, and product. Plus read access to the relevant repositories, pipelines, and cloud accounts, or guided walkthroughs where access is restricted. We agree the exact list, and anything we should not see, in the fit call. Where we process personal data on your behalf, a data processing agreement (DPA) applies.
Answer
What does it cost, and is the assessment fee credited?
Production AI Stack Assessment: Standard from €4,500, Regulated from €8,500. Production Implementation Sprint from €18,000 (4–6 weeks). Private LLM Platform Package from €24,000 (6–8 weeks). Platform Operations Retainer from €6,000/month. Each price is fixed once the scope is agreed. Book a Production Implementation Sprint or the Private LLM Platform Package within 90 days and we credit 100% of the assessment fee.
Answer
Standard vs Regulated assessment: which do I need?
Standard (from €4,500, 10 business days) fits one team and a single primary workload, AI or not. Regulated (from €8,500, 12–15 business days) fits multiple teams, high-risk use cases (EU AI Act Annex III), or an active procurement, NIS2, DORA or regulator review. It adds multi-team interviews, provider vs deployer role mapping, an AI agent and MCP readiness review, and a deeper infrastructure-as-code (IaC) drift assessment.
Answer
How does this relate to NIS2, DORA and the EU AI Act? Is it legal advice?
The assessment maps your systems against the technical measures these reviews ask about: system inventory, logging, access control, incident readiness, documentation, and human oversight. Then it builds the evidence behind them. NIS2 (Network and Information Security Directive 2) has been in force in Germany since 6 December 2025 and covers roughly 29,500 companies. Essential entities must evidence their measures to the BSI (Federal Office for Information Security) by December 2028. DORA (Digital Operational Resilience Act) applies to financial entities. EU AI Act transparency duties under Article 50 apply now. High-risk obligations are a 2027 runway: Annex III high-risk obligations apply from 2 December 2027. All of this is technical evidence, not legal advice. Classification, scoping, and regulatory submissions stay with your counsel, who get a far better starting point from a filled inventory.
Answer
What does EU-controlled infrastructure mean in practice?
Your workloads run inside EU regions or on a sovereign cloud, with keys you hold, open components you can replace, and a documented exit. Options include the EU regions of AWS, Azure, and GCP, the AWS European Sovereign Cloud, STACKIT, IONOS, Open Telekom Cloud, and bare metal. We name these as deployment options, not partnerships. The boundary is defined in Terraform, so a reviewer can read it, and code lives in your repositories from day one.
Answer
Why a fixed price instead of day rates?
Fixed-price scopes are standard in the DACH boutique market. Listed prices are starting points; the price is fixed once we agree the scope, before kickoff. You know cost, timeline, and deliverables up front, and we scope tightly in the fit call so there are no open-ended SOW surprises.
Answer
Why pay for an assessment instead of a free discovery call?
Because you get delivery, not a sales conversation. In 10 business days you receive the full assessment pack (see the 10-day plan above), ending in the straight answer: go, fix-first, or pause. The 30-minute fit call is free; the assessment is where the work starts.
Answer
How is this different from larger AI readiness programmes?
Many readiness programmes run for several weeks across multiple platforms. Ours takes 10 business days and is engineering-grade. You get a 6-dimension maturity heatmap, a FinOps opportunity register, and the system and evidence inventories your reviewers will ask for. Plus a 90-day ranked backlog your team can build from directly. Senior delivery without a programme-sized budget.
Answer
What does the honest verdict mean?
Every assessment ends with a clear recommendation. Go to production with named conditions, fix-first with a ranked backlog, or pause until data, security, or cost foundations are ready. We do not sell a build you do not need.
Answer
What happens after the assessment?
You own the backlog. Implement it internally, or book a Production Implementation Sprint or the Private LLM Platform Package at a fixed price per agreed scope. Add the Platform Operations Retainer later if you want it. No pressure to continue.
Answer
Do you work with our existing data stack?
Yes, that is the default. We work on the warehouse, lakehouse, orchestration, and streaming tools you already run. On top we add data contracts, quality tests in CI, lineage, and freshness alerts. We only recommend replacing a tool when the assessment shows it is blocking production, and we explain why.
Answer
Do you do DevOps and Terraform work outside AI workloads?
Yes. The governed Terraform foundation and DevOps & automation are two of our three practices, not add-ons to AI work. We build CI/CD and GitOps pipelines, progressive delivery with automated rollback, container and Kubernetes platforms, DevSecOps scanning, and observability. We write Terraform/OpenTofu modules with policy-as-code and drift detection for web platforms, SaaS backends, and data platforms, whether or not a model is involved. The same reviews apply: security, procurement, and NIS2 ask the same questions of a web platform as of a model. Non-AI scope is agreed in the free fit call, at the same fixed price per agreed scope.
Answer
We already have DevOps. Why MLOps and LLMOps?
DevOps ships code. MLOps and LLMOps ship models, prompts, and agents with versioning, drift monitoring, evaluation gates, and cost attribution. Production AI needs both, on the same pipelines, so the evidence comes out of one place. We bridge that gap.
Answer
Do we have to self-host models to keep control?
No. The Private LLM Platform Package (from €24,000, 6–8 weeks) is an EU-controlled LLM platform: a gateway with guardrails, audit logging, and cost control. It routes to EU-hosted APIs or to self-hosted open-weight models (vLLM or equivalent). Most teams start with EU-hosted APIs behind the gateway and self-host only the workloads where residency or model control demands it. Choose the Production Implementation Sprint when one workload plus landing-zone modules is enough. Choose the Package when the LLM platform itself is the goal.
Answer
Do you work with our cloud?
Yes. The EU regions of AWS, Azure, and GCP, and sovereign options such as the AWS European Sovereign Cloud, STACKIT, IONOS, and Open Telekom Cloud. Hybrid and bare-metal GPU hosts too. The Private LLM Platform Package ships the same IaC pattern across those targets. We integrate with your existing stack rather than forcing a platform swap.
Answer
Will our team be able to run it after you leave?
That is the point. Everything we build lands in your repositories as Terraform/Pulumi modules, pipelines, and dashboards, with runbooks and an enablement session at handover. Code lives in your repositories from day one. The retainer is optional, not a dependency.