AI does not change work as an abstract future slide. It appears in small moments: a prototype that suddenly comes together faster, a prompt that cannot be reproduced, a workflow that looks different again one week later.
That is why a small circle can be more useful than a large panel. On stage, success is sorted and polished. In a room with people who actually build, unfinished work is allowed to stay unfinished for a while. That is not a flaw. It is the learning space.
Berlin has enough hype. What is missing are honest workshop notes.
The AI Support Group in Berlin-Kreuzberg is meant as a counter-format: no pitch night, no investor filter, no finished answers. Instead there are short impulses, mini Show & Tell, concrete workflows and the question of what people in product, code, design, film, writing, media and culture are actually learning.
Small circles work best when they do not try to look like large events.
The support-group framing is half serious and half ironic. Serious because many people are working through the same open questions: how much automation is useful? What remains human responsibility? Which tools save time, and which only create more work? Ironic because the evening is still allowed to feel light.
Why recaps matter.
A session is over quickly. A good recap makes it useful again later. It does not only record who was there or which tool was shown. It collects the frictions: where did an agent fail, which slide changed the discussion, which question stayed open?
For SEO, this is not a trick. It is the cleanest strategy: real experience, clear context and a repeatable structure. If every session becomes a readable record, a practical archive for AI work in Berlin grows over time.
What a good evening should produce.
- A concrete example instead of a broad AI prediction.
- A workflow that is reproducible, or at least understandable.
- An open question that does not need to be solved immediately.
- An honest failure that saves someone else time.
The blog will keep exactly these things. Not as polished conference documentation, but as working notes from a small circle. If that works, the real value is a local, practical and searchable memory for AI work.