Looking for AI consulting without the bullshit?
AI consulting from MacNorris, a boutique in Hamburg: we find the processes where AI really saves time in mid-sized companies, and build the first solution with you. How long that takes is something we look at together before we start. Not classic consulting: we build with you instead of just recommending.
What is AI consulting?
AI consulting helps companies find out where artificial intelligence delivers measurable value, and introduce those applications. At MacNorris that means: we review processes, data and tools, pick the use case with the biggest lever and build it as a working solution.
According to the German Federal Statistical Office (opens in new tab), around 26 percent of companies in Germany used AI in 2025, and 36 percent of mid-sized companies with 50 to 249 employees. The rest are still waiting. That's exactly where we come in.
A Copilot license is not an AI strategy.
The problem isn't that companies have too little AI. The problem is that nobody knows where to start, and in the meantime, shadow IT grows, data gets duplicated, and tools pile up that nobody maintains. In a few years, the first AI consolidation projects will kick off. We've seen this film before.
Not starting is still not an option. While you wait, your competitors are already running working solutions. A fast way to a first result is a hackathon that produces a prototype in 48 hours.
Are they perfect? No.
Are they better than nothing? Absolutely.
Which one are you?
None of them is a strategy. All of them are avoidable.
The non-starters
Data privacy concerns, uncertainty, too many open questions. Understandable, but meanwhile, the competition is building.
The risk: Those who don't start now will be catching up in two years under twice the pressure.
The action-takers
Copilot here, AI module there, ChatGPT wrapper somewhere else. High costs, little impact. At least they're "doing something".
The risk: Uncontrolled shadow IT and costs with no measurable value.
The overbuilders
Custom AI systems that talk to everything, but nobody understands them, nobody maintains them, nobody knows what they cost.
The risk: Technical debt 2.0, this time with an AI label on it.
No AI deck. No roadmap for Q3. Just the next concrete step!
Our AI consulting does not start with a slide deck. We look at your entire environment: tools, departments, data foundation, already running solutions. We identify the no-brainers, where can AI take over work tomorrow that nobody wants to do manually today? A typical example is AI in sales, where meetings can be prepared in minutes.
And we show you where shadow IT is already growing, before it becomes a problem.
That's exactly what makes us the AI consulting partner that actually works for mid-market companies, no enterprise playbook, just solutions that fit your size. We openly show which AI tools we use ourselves, such as ElevenLabs for AI voices and Higgsfield for AI video.
To make sure your team actually uses AI, we also offer AI training with examples from your daily work.
How AI consulting works with us
Four steps from the first conversation to a solution your team runs itself.
First conversation
You describe where things are stuck. We tell you honestly whether AI helps there or whether the problem lies elsewhere.
Assessment
We look at tools, data and processes and find the no-brainers with the biggest lever.
First solution
We build the first use case and bring it into daily work. For example an AI agent for sales that makes product knowledge plus ERP and CRM data instantly available.
Handover
Your team understands and runs the solution itself. Then you decide whether the next use case follows.
A mini glossary of AI terms
AI terms come at you from everywhere, in meetings, in LinkedIn posts, in every second tool update. We figured we'd collect the important ones in one place, for anyone who's always wanted to know what RAG actually means. If you already know all of this, feel free to keep scrolling.
Prompt: The instruction or question a person gives to the AI. The skill of writing good prompts is called prompt engineering. How that works without a tech degree is covered in prompt engineering for non-technical people.
AI model: The trained system itself, which has learned patterns in data to generate answers, text, or decisions from them. Not magic, just a lot of math.
LLM: The actual language model behind it, GPT, Claude, Gemini, and so on. The foundation your AI solution is built on, but only the foundation, not the finished solution.
Generative AI (GenAI): The umbrella term for AI that can create new content on its own, text, images, music, or code, as opposed to classic AI, which only analyzes or sorts.
AI agent: A model that doesn't just answer, but also acts, calls tools, plans steps, and completes tasks, without you having to be present for every single click.
RAG: Instead of an AI answering purely from its training, it first retrieves relevant information from your own documents and builds the answer on that. That's why it hallucinates less and actually knows your company, not just the open internet.
Context window: How much a model can remember before it forgets the start of a conversation. Small means you have to keep repeating yourself. Large means it keeps track, even across long documents.
Token: The smallest unit of text an AI breaks language into before processing it, sometimes a whole word, sometimes just a syllable. Important for you, more tokens usually means more cost, that's the billing unit behind most AI tools. What that means for your bottom line is in our piece on EBITTDA and AI costs.
Hallucination: A phenomenon where an AI generates false information, but states it so confidently and convincingly that it sounds like the absolute truth.
Fine-tuning: The process of taking an already trained AI model and training it further on specific data, for example your company's internal documents or industry literature, to optimize it for a particular purpose.
Typical challenges in AI consulting
AI Hackathon: Build Real AI Solutions in 48 Hours
AI hackathon: why 48 hours of building together beats 12 months of AI strategy, and how we work with your team to ship real solutions fast.
AI Automation: Start With the Right Process
AI automation: why most projects target the wrong process, how to find the right one, and what a realistic first step actually looks like.
AI Basics: What Must Be Right Before AI
AI basics that need to be right before AI makes sense at all: clean data, clear processes, defined ownership. Why AI rollouts fail without them.
AI Strategy Consulting: From Slides to Results
AI strategy consulting: what a real AI strategy looks like versus a slide deck, and why the most annoying use case is the best starting point.
AI Tools Comparison: ChatGPT, Claude and Gemini
AI tools overview: ChatGPT, Claude and Gemini compared, and why choosing the tool is not the most important question for your AI project.
Vibe Coding: 80% Hype. 20% Fight.
Vibe coding: the first 80 percent build fast. The last 20 percent are a real fight. What the hype videos do not show, and when vibe coding actually works.
Frequently asked questions
AI changes digital office work first: writing, research, analysis, customer communication and recurring processes. Work with atoms changes more slowly, meaning craft, physical work and face-to-face contact. Still, the shift has only just begun. Those who start now help shape it instead of reacting later.
We don't recommend one model across the board, because the choice of LLM rarely makes the difference. ChatGPT, Claude and Gemini are more alike than their marketing admits. What matters is which problem you want to solve and how the AI gets to your data and systems. Starting with the tool question means asking the wrong question.
To get started, the AI features in your existing tools are often enough, but they don't replace a strategy. They save time quickly with little effort. A real AI strategy looks at the full picture: which processes can be automated, what data you have, and where the biggest lever is. That's where we start.
Yes, we do that regularly. We look at your tools, licences and interfaces and tell you honestly what stays, what can go and where AI fits in. We don't earn money on licences, so there's no vendor bias. The result is a stack your team understands and can run on its own.
Now, because experience beats waiting. The market moves fast, but tomorrow's perfect solution never arrives finished. If you start today with a small use case, you learn what works in your company. If you wait, you'll be catching up in two years under twice the pressure.
How long an AI project takes at MacNorris depends on its scope. That's why we look at where you stand in the first conversation and then tell you honestly what to expect. We plan in weeks, not quarters, and start with the use case that has the fastest effect.
Yes, AI consulting and KI-Beratung mean the same thing. AI stands for artificial intelligence, KI is the German abbreviation for the same. In German-speaking mid-sized companies KI is the more common term, internationally it's AI. With us you get the same under both names: finding where AI really helps, and building the solution together.
Yes, our AI consulting is made for mid-sized companies and SMEs. We don't build enterprise solutions that need three departments to understand. We start with a concrete use case that fits your size and your data and build it pragmatically. That way you pay for results instead of complexity.
Not in Hamburg? No problem.
Most of our work runs remote anyway. If you happen to be in Hamburg, we're happy to meet in person too, no fixed rhythm.
No discovery theater.
Tell us where you stand and we'll tell you in one conversation where the no-brainer is for your personal problem.
Further reading
AI Automation: Start With the Right Process
AI automation: why most projects target the wrong process, how to find the right one, and what a realistic first step actually looks like.
AI Hackathon: Build Real AI Solutions in 48 Hours
AI hackathon: why 48 hours of building together beats 12 months of AI strategy, and how we work with your team to ship real solutions fast.