Every offer has tiers (Good / Better / Best) and a clear what / outcome / price / lead time. Pick what fits, or combine into a bundle. All prices EUR, ex-VAT, lead times from signed SOW.
LLM-powered automations that replace repetitive knowledge work.
Email/inbox triage, document classification, data extraction into systems of record, approval routing, report generation.
| Tier | Scope | Price | Lead time |
|---|---|---|---|
| Good | 1 automation, 1 system, ≤2 integrations | €8,000 – €12,000 | 3–4 wks |
| Better | Up to 3 automations, RAG over 1 doc set | €18,000 – €25,000 | 5–7 wks |
| Best | Department-wide, 5+ integrations, agent orchestration | €35,000 – €60,000 | 8–12 wks |
Outcome SLA: defined automation success metrics (e.g. "≥80% of inbound X classified correctly"), measured before handover.
Retrieval-augmented assistant over your own documents (Confluence, SharePoint, file shares, tickets), with citations and access control. Runs on your hardware or in your tenant.
| Tier | Scope | Price | Lead time |
|---|---|---|---|
| Good | 1 doc source, ≤5k docs, chat UI | €9,000 – €13,000 | 3 wks |
| Better | 2–3 sources, SSO, citations, usage analytics | €16,000 – €22,000 | 4–6 wks |
| Best | Multi-team, RBAC, audit log, on-prem inference | €30,000 – €50,000 | 7–10 wks |
Extract structured data from contracts, invoices, claims, forms, IDs, technical drawings — local or cloud models.
| Tier | Scope | Price |
|---|---|---|
| Good | 1 document type, ≥90% field accuracy | €10,000 – €14,000 |
| Better | 3–5 doc types, validation UI, exception queue | €20,000 – €28,000 |
| Best | 10+ doc types, ERP/CRM write-back, human-in-the-loop | €40,000 – €70,000 |
Goal-oriented agents that use tools (APIs, databases, internal systems) to complete multi-step tasks — tier-1 support, procurement research, change-readiness checks.
| Tier | Scope | Price |
|---|---|---|
| Good | 1 agent, ≤3 tools, sandboxed | €12,000 – €16,000 |
| Better | 2–3 agents, prod integrations, evals harness | €22,000 – €32,000 |
| Best | Multi-agent, observability, guardrails, prod ops | €45,000 – €80,000 |
Deflecting tier-1 support over chat/email/WhatsApp/Teams, with human handoff and quality scoring.
| Tier | Scope | Price |
|---|---|---|
| Good | 1 channel, FAQ + knowledge base | €12,000 – €16,000 |
| Better | 2 channels, ticketing integration, CSAT loop | €22,000 – €30,000 |
| Best | Omnichannel, on-prem model, QA dashboard | €40,000 – €65,000 |
Self-hosted inference that runs where your data lives — to SRE standards.
A production-grade self-hosted LLM stack — vLLM/Ollama/llama.cpp, Open WebUI, Traefik TLS, monitoring, backups, runbooks. Runs on NVIDIA, AMD, Intel, or Apple Silicon (Mac). Based on the proven UbuntuLLM Ansible accelerator.
| Tier | Scope | Price (ex-hardware) | Lead time |
|---|---|---|---|
| Good | Single GPU, 1–2 models, basic monitoring | €12,000 – €16,000 | 2–3 wks |
| Better | Multi-GPU, model catalog, SSO, audit log, DR drill | €22,000 – €35,000 | 4–6 wks |
| Best | Multi-node/multi-vendor GPU, HA, ITSM integration, full SRE handover | €45,000 – €80,000 | 8–12 wks |
Hardware: client purchases (we specify); or we procure with a 10–15% markup.
A 1–2 week consulting engagement mapping which workloads can/should move to private AI, the compliance posture, TCO vs cloud, and a roadmap.
| Tier | Scope | Price |
|---|---|---|
| Good | Assessment report only | €5,000 – €7,000 |
| Better | Report + architecture + cost model | €9,000 – €12,000 |
| Best | Report + PoC deployment of one model on client HW | €14,000 – €18,000 |
Cloud + on-prem routing policy (sensitive → local, bursty → cloud), edge inference at site/factory, cost-optimized hardware (Intel XPU, AMD, Apple Silicon) where it fits.
| Tier | Scope | Price |
|---|---|---|
| Good | Routing policy + 1 local model | €10,000 – €14,000 |
| Better | Hybrid gateway, cost telemetry, failover | €20,000 – €28,000 |
| Best | Multi-site edge + central management | €40,000 – €60,000 |
Fine-tune open-weight models on your data, fully on your hardware — no data egress. LoRA/QLoRA, eval harness, version registry.
| Tier | Scope | Price |
|---|---|---|
| Good | 1 task, 1 LoRA, eval report | €10,000 – €14,000 |
| Better | 2–3 tasks, eval harness, regression set | €18,000 – €26,000 |
| Best | Continuous tuning pipeline + registry + rollbacks | €35,000 – €55,000 |
We operate what we built — monitoring, model updates, prompt/model tuning, incident response, and a monthly health report. Keep it running, keep it accurate, extend it over time.
Monitoring, model updates, prompt/model tuning, incident response, monthly health report.
| Tier | Hours/mo | Price/mo |
|---|---|---|
| Good | ~2 days | €3,000 – €4,000 |
| Better | ~4 days | €5,500 – €7,500 |
| Best | ~8 days + on-call | €10,000 – €14,000 |
We run your on-prem inference under SLA, billed per GPU-month. Patching, model catalog updates, capacity advice.
| Tier | SLA | €/GPU/mo |
|---|---|---|
| Good | Business-hours, best-effort | €1,500 – €2,000 |
| Better | Next-business-day, monitored 24×7 | €2,800 – €4,000 |
| Best | 4h critical, 24×7, on-call rotation | €5,000 – €6,500 |
Keep automations accurate as data and prompts drift; add new flows.
| Tier | Scope | Price/mo |
|---|---|---|
| Good | Maintenance + 1 new flow/qtr | €2,500 – €3,500 |
| Better | + evals harness + 1 flow/mo | €4,500 – €6,000 |
| Best | Full automation ops team extension | €8,000 – €12,000 |
Leadership or IT session on private AI, EU AI Act, build-vs-buy, TCO. Output: a 1-page action plan.
Hands-on training for your engineers on running the local-AI stack, prompt engineering, evals, and ops.
Inventory of AI use, risk classification, documentation templates, human-oversight design.
Combine offers and save. Each bundle is a coherent outcome, not a discount pile.
B2 Readiness (Better) + B1 Local-AI (Good)
€22,000 ~ −12%
A1 (Good) + B1 (Good) + 3mo C1 (Good)
€28,000 ~ −15%
A2 (Better) + B1 (Better) + 6mo C2 (Good)
€52,000 ~ −15%
B1 (Better) + A3 (Better) + D3 (Better)
€60,000 ~ −18%