AI Consultant · Netherlands — Amsterdam · Rotterdam · Eindhoven

There are two expensive ways
to get AI wrong.

The first is fear: hold back, watch competitors quietly get faster, and fall behind. The second is waste: buy enterprise AI licenses, run a training day, and watch nothing about how your people work actually change.

There’s a third way: start from how your team actually works, find the repetitive work AI genuinely does better, and embed it there — people in the loop, doing more of what they’re good at. I’ve done it: one small business’s per-product admin went from hours to a review-and-approve step (the case study). A decade of enterprise systems engineering — including NVIDIA GPU infrastructure at scale — means I understand AI from the ground up, not just the sales deck down. Fully English-speaking. No hype.

Let’s understand what AI can do for your business See how I work

// Services

Where do you want to start?

Most businesses need to answer four questions about AI — not all at once, and not always in order. Engagements are scoped to where you actually are. Typical first engagement: the fixed-scope AI Efficiency Review below, then ongoing retainer or fixed-scope work. Day rates from €1,200; fixed-scope quotes available on request.

01 · Assess

How Well Are You Using AI?

Before investing more in AI, it’s worth understanding where you actually stand. A focused 1–2 week engagement: I look at how your business is currently using AI (or not), identify where you’re leaving value on the table and where you may be taking on risk, and deliver a clear picture of what good looks like for your context. A concrete answer to the question most businesses are quietly asking.

AI Audit Risk Review Opportunity Mapping Roadmap

02 · Implement

Getting AI Working Properly

Most AI implementations fail not because the tools are bad, but because they’re the wrong tool, implemented badly, with no real adoption plan. I identify what fits your specific workflows, get it running properly, train your team to use it well, and measure what actually changes. The goal isn’t having AI — it’s working better because of it.

Tool Selection Workflow Integration Training Change Management Measurement

03 · Build

Custom AI for Your Specific Problem

Off-the-shelf AI tools are built for general problems. If yours is specific enough that a generic product won’t solve it, I build. Custom AI agents, LLM-powered tools, and automation pipelines — Python-native, production-ready, integrated with your existing stack. Built for the actual job, not a demo.

Python LLMs AI Agents Automation API Integration Production

04 · Prepare

Is Your Infrastructure Ready?

A tool is only as good as the environment it operates in. If your data, storage, compute, or security posture isn’t ready to support AI workloads, your AI investments will underperform — or fail quietly in ways that are hard to diagnose. I assess and prepare the foundation before you build on top of it.

Infrastructure Assessment GPU Data Pipelines Cloud Security NVIDIA

// Start Here

The AI Efficiency Review

Not sure where to begin? Begin here. A fixed-scope engagement, typically two weeks: I sit inside your actual workflows, find where hours go to repetitive work, and deliver a ranked list of AI opportunities by time saved — including, just as importantly, what not to automate. You end up with a concrete roadmap and the evidence to act on it, before committing to any tooling, licensing, or build work.

Fixed price, scoped to the size of your business — a five-person studio and a fifty-person firm are different jobs, and they shouldn’t pay the same. You’ll have the exact number before we start.

Request a quote for your business →

Specialized support for small businesses and expat founders. Learn about SMB services →

// Featured Projects

AI in practice

Advice is one thing. Here are things built — real examples of what AI enablement, custom agents, and automation work look like in practice.

Case Study · AI Enablement · Workflow Automation

photography-workflow

GitHub → Read the case study →

A fine-art print business was drowning in per-product admin: metadata, model and property release checks, resolution-based print sizing, Stripe pricing for every variant, Google Merchant feeds. An AI pipeline now does all of it — with a human review-and-approve gate where judgment and liability live. Hours per product became minutes. The photography itself stays entirely human work.

hrs→min

admin per product

automated intake steps

legal gates, none skippable

days, build to launch

Workflow Automation AI Agents Human-in-the-Loop Stripe Google Merchant Open Source

Open Source · MCP Server · Python

meta-data-mcp

GitHub → Full write-up →

One MCP server. 76 open-data sources. A plain-text query — “FX rates”, “court rulings”, “earthquakes near Lisbon” — routes automatically to the right dataset. When no match exists in the registry, the server creates a new plugin autonomously and calls it immediately. No per-source install rituals. No manual routing. Built for LLM agents that need real-world data, connected the way agents actually work.

bundled plugins

~330

tools exposed

install step

meta routing tools

Python MCP Protocol AI Agents Autonomous Open Source 76 Data Sources

// About

Derek Linz

Most people who call themselves AI consultants learned about AI from the application layer up: they used the tools, read the blogs, attended the conferences. I came at it from the other direction.

A decade at Nutanix doing kernel-level debugging, building production analytics platforms from scratch, and serving as the global reference for NVIDIA GPU infrastructure across 700+ engineers taught me to understand tools from the inside out — how they actually work, where they break, and what they need to perform. That’s the lens I bring to AI.

It means I can tell you not just which AI tools to use, but why a particular tool fits or doesn’t fit your situation. I can build things that hold up in production, not just in demos. And I can tell you when AI isn’t the right tool for the job — which is sometimes the most useful thing a consultant can say.

I work with businesses across the Netherlands: SMBs taking their first serious look at AI, mid-market teams trying to get more from investments already made, and technical organisations that need the infrastructure sorted before AI tooling makes sense. Native English speaker — a natural fit for international teams and expat-founded businesses.

LinkedIn GitHub Linzalytics

There are two expensive waysto get AI wrong.

Where do you want to start?

The AI Efficiency Review

AI in practice

Derek Linz

Let’s understand what AI can do for your business

There are two expensive ways
to get AI wrong.