AI-made books on Amazon
I tried the "€2,300 a month with children's stories" promise: how I chose the niche, the rules I set myself and the first week.
Every entry is a test with its problem, its method and its result. Filter by whatever you need to solve today.
I tried the "€2,300 a month with children's stories" promise: how I chose the niche, the rules I set myself and the first week.
Three small uses with a quick return before spending on consulting or development.
Client data, GDPR, the AI Act and who owns what AI generates, in practical terms.
From manual paperwork to AI preparing the drafts, without dropping the human review.
Bots, spam, attacks and abuse attempts from day one, and the minimum protections.
I put the promise to the test with today's assistants and measure what it saves me each week.
Nobody buys an automated workflow, they buy a solved problem. What it really takes.
Courses, prompt packs and middleman tools: what this series of articles has been throwing out.
A verdict with numbers, what I'd do differently and what's left of the promise.
A real need solved with a small app made with AI, from the first idea to using it every day.
Ten weeks in numbers: income, expenses, hours and what they say.
Decisions, writing, shopping and work: what got better, what got worse and where I stopped thinking.
Ads, social media and SEO made with AI: what I spent and what came in.
Logo, product photos and posts: what comes out usable and what gives it away.
What setting up the assistant service for shops really cost in money, hours and scares.
I tried the hidden tricks and the perfect prompt. What's left: a few principles that really change the result.
Which business to start given what the internet offers, and the criteria I used to choose.
The MCPs I use every day, how they're installed and the risk of giving it access to everything.
Cost, effort and what I'd do differently after three years with both networks.
How I built it, what went wrong and what it says when it doesn't know the answer.
Where each agent and each model lives, and why there.
What used to mean building workflows I now just ask the AI for. When they still make sense.
What I run on each machine and why, without the operating system wars.
With agents and harnesses around, why I still open a chat window.
What I do on every new machine before putting it to work.
What they do well, what they cost, and what data I give them and what I don't.
What I use each network for and what runs where.
Permissions, mistakes, costs and why you have to keep an eye on them: what has actually happened to me.
An honest tally: what I do with local AI that I wouldn't do in the cloud, and what isn't worth it.
The site takes an afternoon; the domain, deployment, forms and GDPR don't.
What surrounds the model so it's actually useful: interfaces, RAG, connections and scripts.
What it's genuinely useful for, what it does better than a chat and where it falls short.
Ollama, LM Studio, llama.cpp and Unsloth: what I use each one for.
First weeks with my feet on the ground: what it is, what I've achieved and what I haven't.
More than 50 models tested: which I use, which I deleted and what quantisation is.
What happened when I put it into my real work, compared with what I already use.
The machines and GPUs I use to run models at home, and what it would take to get started.
ChatGPT, Claude, Gemini, Copilot, DeepSeek, GLM, Kimi and MiniMax, and where the limit lies.
What made me switch, what I miss and what I still use OpenAI for.
The setup behind everything I write about on this blog.
Try removing one. If you think this case should exist, you can suggest it.
Clear filtersOne email a week with what was verified, what was refuted and what is still on the bench. No news roundups.