Skip to content

AI specialist for Dutch SMB

We build and deploy AI where it provably removes manual work.

We work with leadership and IT teams of mid-sized organisations who take AI seriously, not as a gimmick.

Who we are

Vectel is a broad IT partner for Dutch SMBs. AI automation is our most substantial practice: we pick the right tool for the problem and are honest when AI is not the answer.

Our AI practice

Five ways we deploy AI, all with a measurable outcome.

  • Microsoft Copilot implementation and adoption

    Roll out Copilot, set up governance, and make sure teams actually use it.

    Book a Copilot conversation
  • AI automation

    Speed up repetitive processes, unstructured data handling, or decision-making with LLMs and agents.

    View the automation practice
  • AI integration in existing systems

    LLM capabilities inside your CRM, ERP, or customer portal, with attention to latency, cost, and governance.

    Discuss the integration
  • AI audit and readiness

    An objective view of where AI delivers value in your organisation and what needs to be sorted first.

    Start the scan
  • AI training for teams

    Practical sessions where teams learn to use AI responsibly in daily work.

    Plan a training

Our approach

Five phases, no consultancy jargon.

  1. Step 1
    Discovery (1 to 2 weeks)

    What problem are we solving, for whom, and how do we measure it?

  2. Step 2
    Scope (1 week)

    What is out of scope, what data do we need, what risks do we accept?

  3. Step 3
    Pilot (4 to 8 weeks)

    Small, measurable, with time to adjust before we scale.

  4. Step 4
    Measure (ongoing)

    What changes, for whom, how much manual work actually disappears?

  5. Step 5
    Scale

    Only when the pilot delivers evidence, not before.

Are you ready?

Five criteria we assess before starting a project.

  • Data quality

    Is the data findable, clean, and do you know who owns it?

  • Governance

    Who can do what, what is the GDPR legal basis, and in what context is it deployed?

  • Leadership

    Is this supported by the board or management team, or is IT on its own?

  • Use case clarity

    Concrete problem or 'something with AI'? The second is a red flag.

  • Integration context

    Which systems need to connect, and on what timeline?

Take the AI readiness scan

Common mistakes

  • AI on messy data

    We typically see that data needs to be sorted first, otherwise AI performs at best the same as the old process.

  • Choosing tooling before the use case

    The tool choice should follow from the use case, not the other way around.

  • Pilot without success criteria

    Without agreed-upon metrics up front, you will not know afterwards whether it worked.

  • Adoption as an afterthought

    A great AI feature that nobody uses is not a win.

  • Compliance at the end

    Bringing in GDPR and NIS2 only at the last moment costs more than addressing them upfront.

Tooling choices

We choose tooling after discovery, never before.

Microsoft Copilot
When the organisation already runs M365 and the use case fits within Office.
Claude or GPT via API
When custom work or integration outside M365 is needed.
Custom agent framework
For multi-step processes with domain logic and autonomous steps.
No AI
When a rule engine, form adjustment, or better search is the real answer.

Data quality and governance

AI only delivers value when data is in order and governance is clear. For organisations under NIS2, that starts with a basic inventory.

Take the NIS2 scan

Cases from our practice

Anonymous examples, because we do not discuss clients without permission.

  • Document extraction

    Invoices, receipts, contracts and email attachments parsed automatically.

    Purchase invoices and receipts retyped into the accounting package eat hours a month and leave typo errors that have to be tracked down later. An AI extractor reads the structured fields, checks them against the order line, and posts them through when the match holds.

    Read more
  • Mail triage for the shared inbox

    info@ and support@ sorted, summarised, routed. A human decides what goes out.

    The shared inbox is the quiet pain of many offices. Three people read the same mail, nobody picks it up because they think someone else will, and the customer waits a day for something that could have been done in five minutes.

    Read more
  • Knowledge portal over your own documents

    One search box over SharePoint, handbooks and wiki. Answer with a source link or an honest "no idea".

    New joiners ask the same questions for three months that are already answered in a document nobody can find. An internal search that knows which page it does not know is worth more than a chatbot that guesses.

    Read more
  • Quote flow for sales

    From a customer request to a draft quote in minutes, not days. A human hits send.

    A salesperson gets three requests in on Monday. By Thursday they are still sitting there, because copying from earlier quotes, looking up price lists, and polishing the branding is the work that gets pushed aside between customer calls.

    Read more
  • Meeting summaries that actually exist

    The kind of meeting whose minutes never get written, now gets a finish line.

    In most SMB meetings nobody writes minutes. Not from ill will, but because the person who would type them is part of the conversation. A short summary afterwards is worth more than a forty-minute transcript nobody reads.

    Read more
  • Lead classification with reasoning

    A readable priority list per lead, with arguments, not a black-box score.

    A lead form on the site, a whitepaper download, a conference badge scan. At the end of the week eighty names sit in the CRM. Which five should sales call today, and why?

    Read more

Outcomes are indicative and project-specific.

Where we operate

We work from Veenendaal for clients across the whole of central Netherlands.

Frequently asked questions

  • What does an AI project cost?

    Depends on scope and data state. After discovery we give a substantiated range, not a fixed price upfront.

  • How long does an AI implementation take?

    A typical pilot runs 4 to 8 weeks. Scaling after that depends on scope.

  • Does AI work for small businesses too?

    Yes, provided the use case is sharp. Being small is no blocker, vagueness is.

  • What if our data is not clean?

    Then the project starts with data, not with AI. Otherwise you get an expensive parrot.

  • Which tools do you use?

    Depends on the use case. We work with Microsoft Copilot, Claude, GPT, and custom agent frameworks where appropriate.

  • What GDPR risks does AI introduce?

    Mainly around purpose limitation, legal basis, and transfers. We address this in the scope phase, not at the end.

  • Do you work together with our current IT partner?

    Yes, often. Our role is then AI specialist within your existing IT architecture.

Ready to start?

Three ways, pick what fits.