Three routes onto one decision, and what separates them is not the input but how much the price can be trusted. A slab you photographed at a yard and a quote a supplier sent you are not the same kind of number, and the screen that pretends otherwise is the one that gets somebody hurt.
Two are already on this decision. Whatever you add sits beside them until you settle it.
Saving something you like without pricing it is the other door, on the kitchen's board. Nothing there has to become an option.
Mockup 44 draws the first: photograph a tap you liked, and it goes to the room you are standing in with no price and no obligation. This screen is the second, and it starts from a decision rather than from a room. Folding them into one flow would put a price field in front of somebody who is only saving a picture, which is exactly where people stop capturing.
Because you opened it from the counter material and the eyebrow says so. A picker here would be a picker whose answer the product already had, and every one of those is a tax on the person who did the right thing by starting in the right place.
This one will say you typed this wherever it appears. That's not a warning about you, it's the difference between this and the quote below it.
An option with no price is still a real option and gets added without one — it simply reads no price yet on the decision, which is honest and is a state somebody can act on. What can never happen is a price with no account of where it came from. That is the whole of G4 stated at the smallest scale the product has.
Rung 1, in the user's own words, and it says what the consequence is rather than that something is wrong. Sam typing a number they worked out themselves is the ordinary case and frequently the most accurate one in the product. The tint marks that the number has one source and it is a person, which matters at the moment three of these sit side by side.
The same shape as recording something a person told you on 23: the machine produced the reading, so a human stands behind it before it becomes a fact. The confirm step is not ceremony here for the same reason it is not there — there is a genuine authority gap, and it is between a photograph and a price somebody will spend against.
Let me correct it sits at the same level as accepting, not behind a pencil somewhere. When a product is confident enough to read a number off a photograph it owes the reader an equally easy way to say no, and burying that behind an edit affordance is how a misread tag becomes the price of a kitchen.
The document stays whole and openable. What we read out of it is a convenience, and the thing that settles an argument is the file.
Which is the same rule the trust ledger runs on and the same one 23 applies to a text: keep what actually arrived, beside the tidied version, because a year later the question is what the supplier actually said and a paraphrase is not an answer to it. Nothing in this product replaces a document with its own summary of it.
It is drawn as a date because that is what it is. The moment it becomes expires in 6 days it is pressure applied to somebody choosing a kitchen counter, and this product does not do that even when the deadline is real and somebody else set it.
Correcting a value corrects what HouseChalk shows. The PDF stays exactly as Meridian sent it, and it is still the thing that settles an argument.
Once you have corrected one, the row says you set this rather than we read this, and it stays that way if the file is ever read again.
Read off page 2, under Total. The reader is not learning how to fill in a field; they are checking a claim, and the only useful thing the screen can tell them is where to look to check it. A hint reading enter an amount would be the format rule forms.css forbids, one layer up.
A corrected value is a decision somebody made, and re-running the extractor over the same file later must not quietly undo it. So the row changes what it says about itself — you set this — and that flag outranks anything a later pass produces.
Not a general edit screen for the option. Everything on it is something the product asserted about a document, which is exactly the set a person might want to dispute. The name, the room and the price the option ends up carrying are edited where the option lives.
Three on the table. Nothing is settled and nothing is recommended.
Whichever you settle, the price and where it came from go onto the decision and stay there. That is what the money tab reads later when it says what a thing cost and why.
A badge is a status and would compete with the number it is qualifying; this is a qualifier on the number and sits in the metadata ramp under it, the way from QuickBooks, read this morning does on the builder's job screen. Same rule, same shape, two surfaces apart.
The product has no view about which counter somebody should have, and a best value chip here would be one invented out of two numbers it does not trust equally. So the list is ordered by price, which is at least an honest ordering, and the sentence underneath says plainly what that ordering does not capture.
The same argument at a bigger number: three quotes for one package, compared on 18. What differs is that a cabinet package arrives as three quotes for the same scope and a counter arrives as three different materials, so one is a comparison and the other is a choice. They share the provenance rule and nothing else.
Three kinds, strongest first. The list itself is ordered by price, because that is the one ordering that is not a judgment. This is what it cannot tell you.
A firm, a date and a document. Porcelain is $9,400 because Meridian Tile & Stone wrote it down on 2 June, and the quote is in your binder next to the number.
A tag you photographed, a page you pasted. We read a number off it and you confirmed we read it right. Honed quartz is $11,800 that way.
You typed it. Soapstone is $14,200 because that is your estimate, and the product treats it as exactly that for as long as it stands.
Average them, mark one as the best value, or pick one for you. The three are not a spread of the same thing, and the cheapest here is the only one anybody has actually promised you. That is a fact worth having, not a recommendation.
No rows, no prices, no action. It exists because the frame before it makes a claim and then declines to act on it: the list stays in price order, and the reader is told that order is not the whole story without being told what the rest of it is. This is the rest of it.
A quote can expire; a checked tag is not an offer; a typed figure never firms up on its own. Those three sentences are the load-bearing ones, and each sits under the tier it qualifies rather than gathered into a disclaimer at the bottom where nobody reads them.
Average them, mark one as the best value, or pick one for you. Same shape as the refusal on mockup 14 and on frame 5 — a sentence on the ground under a section header, since the un-carding pass of 2026-09-06 took the box off all three: the thing the product will not do, said as a decision rather than a gap, because all three are features somebody will propose and a best value chip is the one that looks most like a kindness.