Status: INPUTS RECORDED, DECISION OPEN. The paper design (the spec) is signed off; this document records the look-and-feel discussion that followed the same evening. Per Aristide's gate (2026-08-17: "I want that research in before we start designing"), no interface design or mockups happen until the operator love/hate research is delivered — research launched 2026-08-18, running. The two panel proposals below are discussion material Aristide explicitly asked for, not a design.
After sign-off Aristide asked for three interface directions to look at, discussion before action. Three were offered — same engine, same eight stages and gates under all three; they differed only in what the home surface is:
Jordan then described his own picture, which supersedes all three as the working direction (verbatim, 2026-08-17):
"What I imagine is the CAD view as the main part of the screen, with our toolbars on the left column, then the workflow tabs across the top - left to right - import, heal, QC, review, nest, cut, plan, post, deliver. Think workflow, as in go high level. Need to keep everything as visual as possible. I do imagine that when you switch users, we keep track of where they keep their tools and their favorite tool layouts. Maybe an essential preset like how Adobe does."
Read back and agreed as: the drawing owns the screen on every tab, customer's original as the toggleable gray base layer; eight tabs across the top wearing status colours (green done / yellow flagged / red blocked) so the job's state reads off the tab strip from across the room; the left toolbar's contents follow the active tab; checks live ON the geometry (open contour glows red, bridges and tabs are grabbable markers), numbers behind them in a drawer/panel; per-user remembered tool layouts with Adobe-style presets. This is the Bench's canvas with the Rail's spine turned into top tabs.
1. Where a job starts — Jobs home, folder browse, AND draw-from-scratch (ruling). Aristide accepted the Jobs-home front door (every active job, its tab-strip state in miniature, who has it; double-click lands in the canvas; sales' natural home — the Ledger reduced to a front door). Then he widened intake (verbatim): "In our case it would be a folder that we browse to but it needs to be flexible for workflows for other shops. My customer says I don't have a print. I don't have anything at all. Can you just give me a 2-inch OD with a 0.5-inch ID? Can I just start with a blank document and say, 'Just create two circles and define their dimensions and then move on with my workflow,' or do we have to start with a DXF or something? I want to be able to draw from scratch."
2. The numbers/check-results drawer — position open; he asked for proposals (question, proposals below). Verbatim: "If we do Adobe style, it would be a draggable window across the screen so I can decide how my real estate works. If I have an extra screen, I could even put the toolboxes on the extra screen. However if you see different options, I'd like to see them. Give me a couple of proposals."
3. Canvas-first with the eight top tabs — confirmed. Both of Aristide's questions build on Jordan's picture as the given ("Even with canvas-first, something has to open before geometry exists"); no part of it was contested. Recorded as the accepted direction the look-and-feel design starts from.
And the gate (verbatim, same evening): "I do want you to also spend some time next on that research on what they love and what they hate. That's going to be important. I want that research in before we start designing."
Proposal A — docked slots, tear-off allowed ("the workbench"). Every panel lives in a dock slot: left column (tools), right edge (the checks/ numbers drawer), bottom strip (job log / cut order timeline). Panels drag between slots and collapse to an icon strip; a panel can be TORN OFF into a floating window — which is how the second monitor works (tear off the toolboxes or the drawer, drop them on the other screen) — but on one screen everything not floating snaps back to a slot, so a panel can never be lost. The drawer defaults to the RIGHT edge: check results are vertical lists (parts, contours, flags) and read naturally beside the geometry, and on a wide sheet (60x120 nests) the bottom of the screen is where the canvas needs its pixels. Per-user workspace = which panels sit in which slots, plus any floats. Cheaper to build, consistent across workstations, novice-proof; the ceiling on real-estate control is the slot system itself.
Proposal B — free-floating Adobe workspaces ("the studio"). Every panel is a draggable, snappable window anywhere on any monitor; panels dock into each other as tabbed clusters; named per-user workspaces (a Heal layout, a Nest layout...) with shipped Essentials presets per stage, and Window > Reset Workspace as the escape hatch. Aristide's extra-screen case is native, real-estate control is total, and it matches the Adobe muscle memory Jordan named. Costs: the heaviest UI plumbing of the options, and the classic failure mode is a shared-workstation shop where a panel "disappears" for the next login — mitigated by per-user workspaces + reset, but the research should tell us how much operators actually curse floating-panel UIs on the shop floor.
Suggested middle (for the discussion): A's slots as the base with B's tear-off and per-user named presets grafted on — second-monitor support and Adobe-style workspaces without full floating chaos. The love/hate research lands before this is decided; lost-panel and layout-amnesia complaints are exactly the kind of signal it should surface.
The signed-off paper stays the spec for the engine and stages. The look-and-feel layer starts only after: (1) the love/hate research is delivered (Aristide's gate), (2) the panel-layout choice is made from the proposals above, (3) draw-from-scratch is scoped. Nothing here changes the paper's stages, gates, roles or tables.