Every example has to show what it started from

You are here to find one thing: a piece of work close enough to your own that you can judge whether Studio is worth your time. A kitchen elevation like the ones your team draws. A customer's room with a floor surface in it. A scene that was unfinished and had to be resolved.

That is a reasonable thing to want, and it is the thing this page is for. It is also the thing this page cannot give you yet, and the rest of the page says so plainly rather than filling the space with pictures that would not answer the question.

Show someone a finished interior visual and they still know nothing that matters. They cannot tell what went in. They cannot tell how much of what they are looking at came from a drawing and how much was supplied by the model. They cannot tell how many attempts it took, or how much a person adjusted afterwards, or whether the same input would land anywhere near it a second time.

So a beautiful output oversells and a plain one undersells, and neither tells an evaluator what they need. The fix is not better pictures. It is publishing the things around the picture: the source, the instruction, the workflow, the preserved facts, the inferred ones, and the limits.

These categories are not interchangeable and they are not blurred. A demonstration produced by us does not become a case study because the material looked realistic, and a case study is not published because a customer said something encouraging in a meeting — it needs their permission and an honest account of the parts that did not go well.

Four of the five are evidence. The fifth, explanatory schematic, deliberately is not: it exists so that a diagram of a process carries a label that claims nothing, instead of borrowing a demonstration label it cannot support. A figure carrying it proves nothing, and nothing on this site is allowed to lean on one.

Case studies are kept separate from demonstrations on this page for that reason. When both exist, they will be shown in different groups with their labels visible, so that nobody has to work out which kind of evidence they are looking at.

Nothing is published on this page yet, and the reason is the standard above rather than a shortage of output. An example goes public once its provenance is recorded, its permission is in place, and there is a working route a visitor can use to run the same job. None of the three is in place today.

That also means this hub is not comprehensive. It becomes a fair picture of what Studio does once each flagship workflow has at least two examples behind it, and the copy on this page will change when that is true. Until then, the three commitments below are what the workflow pages have already promised, restated here in their own words so that each promise has a home you can hold it to. Two of them now have an address as well as a wording: the pages linked underneath set out the full evidence sequence each example will have to walk, with every figure on them still empty.

It will not show you an output on its own. Every published visual arrives with the material it started from, at a size where you can compare the two, because a result without its source is decoration.

It will not show one lucky run and let it stand for the rest. Where a result took several attempts, or where a person selected between outputs, or edited afterwards, that will be disclosed on the card. An example that needed help is more useful to you than one that pretends it did not.

It will not show a customer's drawings, rooms or products without their permission, and permission means permission to publish, not general goodwill. That is why case studies are rarer here than demonstrations, and why the two are never mixed together.

It will not name a run time, a throughput figure or a saving. Measured performance is published as a product claim when it is stable enough to stand as one, and not a day earlier.