Open-source AI: what is actually open?
Open source, open weights and self-hosting mean different things. The licence, data flows and maintenance matter more than the label.
Read articleOOG OP AI
One eye on AI.
The other on common sense.
Developments, explainers and practical tips. What can you do with the technology, what still needs checking and when is a simpler solution enough?
10 articles
AI and AGI: why the distinction remains difficult
AGI sounds like a single milestone. Definitions, testing and human assistance make the picture more complicated.
Read articleDeerFlow: what can an AI agent do for your work?
DeerFlow coordinates agents that carry out multiple steps. Permissions, costs and oversight deserve as much attention as the demo.
Read articleMiroFish simulates reactions. Does it predict the future?
AI characters can explore possible responses. The challenge is distinguishing useful questions from false confidence.
Read articleBuilding an application with AI: start with one task
A small demo makes an idea tangible. Keep the first version focused and recognise what is still needed for real users.
Read articleYour own AI chat: where does the data go?
A chat on your own domain does not automatically process everything locally. Look at the models, documents and connections behind it.
Read articleWrite useful AI instructions without prompt tricks
A clear goal, relevant context and boundaries are more useful than a complicated prompt. Includes an example for a customer email.
Read articleHow I check an AI answer
A working source link and a neat table do not prove a claim. Check the facts that determine what you will do with the answer.
Read articleWhat is GitHub, and what do all those stars mean?
You do not need to be a programmer to assess an AI project. Start with its explanation, installation requirements and maintenance.
Read articleStart with the problem. Then consider AI.
Retyping the same data? Find out why it happens before choosing what to automate.
Read article