The demo is not the system
The claim
You have seen the demo. It worked. They all work.
A demo is an artefact with nothing behind it. Somebody prompted a model, kept the output that looked good, and showed you that one. There is no identity behind it: no plan it came from, no repository it landed in, no board that tracked it, and nobody who can tell you what it cost or what it left out. The distance between that artefact and a system your company can run is not a shortcoming of the demo. It is the entire job.
This film spends most of two minutes on what came out before it stops to name the product, and the name card admits you have been looking at it the whole time. Judge the output before you hear the pitch.
Reading the film
0:08 — Eighty three architecture drawings, for the forty one phases of five builds. The caption is on the frame and it carries its own denominator, which is what makes it a rate rather than a boast: about two drawings a phase, every one produced from a plan before the code existed. A drawing is either consistent with the system it describes or it is not, so it is the cheapest claim in the film to falsify.
0:39 — One Word file a client sent. A canonical product vision, with a table of contents and a roadmap, written for people rather than for a machine. This is the actual input. Nothing about it was prepared for this film, which is why it is worth more than a prompt would be.
1:03 — It found three repositories named inside that prose. The panel reads “Detected 3 repos in the document and pre-selected them below”, and beneath it every feature is already assigned to the repository responsible for it. Nobody told it there were three. That is the difference between reading a document and accepting a form.
1:15 — Twelve phases. Seventy work units. Five hundred and sixty hours. The hours are added up from the units rather than guessed at the top. That distinction is the whole difference between an estimate and a number: you can open any unit and see what it contributed. This is the twelve-phase plan the Unboxing takes apart, phase by phase, in the next article.
1:35 — Two hundred and ten gaps, twenty one of them in phase one. A gap is a question the plan cannot answer on its own. It does not fill that silence. This is the beat to watch if you are trying to decide whether the thing is honest, because filling those gaps convincingly would have been easy and would have looked better.
1:51 — The name, and what it admits. StellarView. You have been watching it since the first frame. Everything above happened before you were told what made it, and that order is the argument.
2:22 to 2:43 — Your JIRA, your repository, your wiki. Three beats, one point. A ticket auto-created from a support ticket and linked back to it, a merged pull request with its own safety case, and a company wiki under the customer’s name. Nothing above required you to adopt a new place to work, and that is the claim article six has to answer for.
What this costs, and what it does not do
The film shows outputs, not the work producing them. You are watching artefacts and a narration, not a live run, and the five hundred and sixty hours is a plan’s estimate rather than a measured actual. The two hundred and ten gaps are real work left for people. Nothing here proves the plan was good, only that it exists, is itemised, and refuses to answer what it cannot.
That looked like a lot of output for one document. Ask what got skipped.
The SDLC didn't go away →See it running
sue-torch in the Stellarverse index →The same film, next to every other one, with the evidence for each sitting under the film that pays for it. No price and no form on that site.