Home / AI customer service
AI customer service · Netherlands

AI customer service that deflects tickets, not customers

AI chatbots and support agents that answer instantly, resolve the repetitive questions, and escalate the rest cleanly to your human team — grounded in your own content.

Last updated: 11 June 2026

Summarize with AI Prompt copied — paste it into the chat
Share
In short

AI customer service uses chatbots and AI agents to answer common questions instantly, deflect repetitive tickets, and escalate the rest to humans — grounded in your own content so it does not invent answers. Crux Digits ships it at fixed prices: a €2,500 audit, a €20,000 proof of concept, and production from €50,000.

What AI customer service actually does

It is not a bigger FAQ page. A well-built system reads the question, finds the answer in your own knowledge base, and either resolves it or hands off with full context — so your team spends its time on the tickets that genuinely need a human.

Chatbot or AI agent: which do you need?

Both have their place. A retrieval chatbot answers questions; an AI agent also takes action. We help you pick the right one for the job — and often the answer is a chatbot first, an agent later.

DimensionRetrieval chatbotAI support agent
Best forAnswering questions from your contentCompleting tasks: lookups, updates, actions
BehaviourFinds and explains the right answerReads context, then acts in your systems
IntegrationsKnowledge base, docs, websiteCRM, order system, ticketing, internal APIs
Risk profileLower — it informs, humans actHigher — needs guardrails and approval steps
Start here ifMost of your tickets are repeat questionsYou want to automate whole workflows end-to-end

Grounded in your content, so it does not hallucinate

The difference between a demo and a system you can trust is grounding. We build on retrieval-augmented generation (RAG): every answer is pulled from your documents, policies and product data, with citations back to the source. If the answer is not in your content, the assistant says so and escalates — it does not guess.

Everything is built EU AI Act-ready and AVG/GDPR-proof: your data stays under your control, answers are traceable, and you own the code. No black box, no lock-in, no customer data leaking into someone else's training set.

What AI customer service costs

No open-ended retainers and no surprise invoices. We work in three fixed-price bands, so you know the number before we start:

  • AI-readiness audit — €2,500. We map your ticket volume, content and systems, and tell you honestly what to automate first.
  • Proof of concept — €20,000. A working assistant on your real content, so you can judge it on your own tickets before scaling.
  • Production — from €50,000. The hardened, integrated system, live for your customers and owned by you.

How we ship it: audit → PoC → production

We de-risk it in stages instead of asking you to buy a big platform on faith. The audit tells you what is worth automating. The proof of concept proves it on your real content and tickets. Production hardens it, integrates it with your ticketing and CRM, and hands you a system you own outright. You can stop after any stage — but most clients do not want to.

FAQ

Frequently asked questions

What does AI customer service cost?

Fixed prices: €2,500 audit, €20,000 proof of concept, production from €50,000. You know the total before we start, with no open-ended retainer. The audit scopes your content and ticket volume; the proof of concept runs on your real knowledge base rather than a demo set.

  • All prices exclude VAT. Most Dutch AI agencies publish none, so comparing them usually means a sales call first.
  • Running cost after launch is model usage plus hosting — normally a small fraction of the build.
  • You own the code and models, so there is no per-seat licence and no lock-in.

Should I use a chatbot or an AI agent?

Start with a chatbot if most tickets are repeat questions. A retrieval chatbot answers from your content. An AI agent goes further and takes action in your systems — order lookups, status updates, refunds. Agents carry more risk, so they need guardrails and approval steps.

  • Chatbot integrations: knowledge base, docs, website. Agent integrations: CRM, order system, ticketing, internal APIs.
  • Most teams start with a chatbot and add agent capabilities once answer quality is proven.
  • Rule of thumb: a chatbot informs and humans act; an agent acts, so it needs an approval step.

Does it work in both Dutch and English?

Yes. It answers each customer in the language they wrote in, from the same knowledge base. Dutch and English are both handled natively, and other languages work too. You do not need a separate bot or a translated knowledge base per language.

  • One source of truth: your Dutch handbook can answer an English question, and the reverse.
  • Worth testing during the proof of concept — quality varies more on your jargon than on the language.
  • Crux Digits works bilingually in Dutch and English, so both versions get reviewed by a person.

Will it hallucinate or invent answers?

It is built not to. Every answer comes from your content via retrieval-augmented generation, with a citation to the source. When the answer is not in your knowledge base, it says so and escalates to a human instead of guessing. You can inspect that behaviour in the code.

  • The refusal policy is the thing to test — ask any supplier to demo a question their bot cannot answer.
  • Citations make each answer auditable, which matters most for policy, pricing and warranty questions.
  • Grounding removes most hallucination but not all, so human escalation is the backstop rather than an optional extra.

How long does it take to launch?

The audit takes 1–2 weeks; a proof of concept 4–6 weeks. Production runs from 8–12 weeks and depends on the integrations — ticketing, CRM, internal systems — which get scoped precisely during the audit. That scoping is what stops the later timeline from slipping.

  • The proof of concept runs on your real content, so you see genuine answer quality before production.
  • Integration count is the main timeline driver: a website-only bot is far faster than one wired into ticketing and CRM.
  • You can stop after any stage, and each one ends with a deliverable you keep.

See what AI could deflect from your support queue

Start with a €2,500 audit: a clear, honest read on what to automate first, what it will cost, and what it will return. No open-ended retainers, and you own everything we build.

Book your AI-readiness audit