The Canvas
Navigate, arrange, and organize your infinite thinking space.
Pan, zoom, and navigate
The canvas is an infinite space. Click and drag the background to pan, or scroll with two fingers on a trackpad. Pinch to zoom, or hold Cmd (Mac) / Ctrl (Windows, Linux) while scrolling to zoom toward your cursor. On a Mac, holding the Control key gives you an eagle-eye view — the canvas zooms out to show everything, and releasing dives back in at your mouse position (tune it under canvas settings). Subtle concentric rings around the canvas center, with a small crosshair at the origin, help you gauge how far you've panned — and how far you've zoomed: ring spacing stays fixed in the world, so sparse rings mean you're zoomed in close and dense faint rings mean you're far out. Once you have at least one thought, panning gently stops at the edge of your thoughts — you can push the outermost node almost to the edge of the screen but never so far that everything leaves the view, so you won't get lost in empty space. If a zoom nudges your thoughts past that edge, the canvas glides back to bring them into view instead of snapping.
| Shortcut | Action |
|---|---|
| Cmd/Ctrl + Scroll / Pinch | Zoom in or out |
| Scroll / Click + Drag background | Pan the canvas |
| F | Fit all content in view |
| Cmd/Ctrl + / Cmd/Ctrl - | Zoom in / out |
| Hold Control (Mac only) | Eagle eye — see everything, release to dive back in |
Selecting nodes
Click a node to select it. Hold Shift and click to add more nodes to your selection. Drag a rectangle on the background to box-select multiple nodes. Use Cmd+A (Ctrl+A on Windows) to select everything. Two different toolbars appear depending on how you selected: clicking a single node floats that node's own tools above it — open as garden, edit, select children, open in panel (which of them appear depends on the zoom level and the node) — and pops a + handle out of each side of the node (left = parent side, right = child side; click to add, drag to connect — see Connections & Edges), which is where adding a thought lives now — while a box-drag (or any selection of two or more) brings up the group toolbar beneath the selection, with fit, align, tidy, lock and delete, plus the same pair of side handles on the selection box for connecting everything you selected in one pull. Only one toolbar is ever on screen: a box drawn around a single node hands it to the group toolbar and that node's own tools stand down, while clicking the same node does the reverse. Deleting, locking and fit-to-selection live on the group toolbar, so reach them by drawing a box — even around a single node. On a keyboard, Delete removes the selection outright and right-click offers the same actions.
| Shortcut | Action |
|---|---|
| Click | Select a node |
| Shift + Click | Add to selection |
| Drag background | Box select |
| Cmd/Ctrl + A | Select all |
Undo and redo
Nearly every canvas action can be undone: creating, renaming, and editing thoughts (from the canvas editor, the low-zoom inline title, or the side panel), deleting thoughts, moving one node or a whole multi-selected group, duplicating and placing thoughts, nesting and un-nesting, drawing/deleting/flipping/relabeling connections, aligning nodes, arranging a selection (Tidy Up, or Outline, Grid, Force-directed, Tree and Radial from the selection toolbar's Arrange menu), color-region changes (add, move, rename, recolor, delete), and actions performed by the AI assistant (created thoughts, edits, moves, connections). Use Cmd+Z / Cmd+Shift+Z (Ctrl on Windows, plus Ctrl+Y for redo), the toolbar's undo/redo buttons, or the > Undo / > Redo commands. The history is scoped to the canvas you're looking at — switching gardens or drilling into a nested canvas starts a fresh history. A few things intentionally stay off the undo stack: accepting or dismissing AI suggestions (dismiss IS the undo), layout-mode switches (choose Layout: Manual to restore your hand-placed positions), and permanent deletes from Trash. If an undo fails (for example a network hiccup), it stays available so you can simply try again.
| Shortcut | Action |
|---|---|
| Cmd/Ctrl + Z | Undo |
| Cmd/Ctrl + Shift + Z | Redo |
Locking placement
When you box-select nodes (or select two or more), the group toolbar that floats beneath them includes a lock button (between Tidy Up and Delete). Lock freezes the selected nodes' position so a stray drag won't move them, and Align, Tidy Up, and the other arrange tools skip them. Locked nodes render slightly dimmer so you can tell at a glance what's pinned — and everything else (selecting, editing, connecting) keeps working. While locked, the button shows a blue open-lock icon — click it to release them. Locks are remembered per node on this device.
Creating thoughts
Double-click (or double-tap on mobile) any empty space on the canvas to create a new thought. When zoomed out, a lightweight inline title input appears right at the current zoom level — no need to zoom in first. When zoomed in, the full editor opens. You can also press Cmd+N or use the command bar (Cmd+K) and type a name.
Level of detail
Nodes adapt their appearance based on zoom level, like a map moving between street and country views. When zoomed in (above 70%), you see the full card. The full card itself has two stages: near the threshold it opens condensed — title and content, with small counts showing how many nested thoughts and links it carries — and its properties, nested-children, and backlinks sections unfold as you zoom in past 80% — a little sooner on a phone, where the canvas stops zooming in earlier, so the fully expanded card is always reachable. Zooming out steps down through progressively lighter views: a compact card (title, snippet, and the same counts), then a chip (title with child and connection counts), then a dot with a label, and finally a minimal colored dot at the furthest zoom. Long titles stay readable: the full card always shows the complete title, the compact card and chip give it two lines before trimming, and hovering any node whose title is cut off reveals the whole thing in a tooltip. On desktop these views morph smoothly into one another as you zoom — the marker and title hold their place while the surrounding card fades in or out; on phones the views switch in discrete steps. Each view has one consistent size; labels brighten with each level up and grow when a thought moves from map-style labels to cards — so next to each other, a more detailed thought reads as more important, the way country labels outweigh city labels on a map. Thoughts that carry a picture — a generated image, an Album or Track cover, or the first image in a description — keep it at every view: the full card opens with a wide cover above the title, the compact card and the chip show a small thumbnail where the type icon would be, and both dot views draw a tiny picture marker instead of the plain colored dot, so you can still spot an image on the map from far away. In those smaller views a small colored dot on the picture's corner keeps the thought's type visible, and for generated images, hovering the picture shows the prompt that made it. Only the very deepest zoom simplifies back to plain colored dots. At the deepest zoom most dot labels fade away to keep the overview clean, but hovering over or selecting any dot fades its name back in, and dots with connected children keep their name and collapse chevron visible. When a dot's name is showing, the name itself is clickable — you don't have to hit the dot. No thought ever disappears — every node stays on the canvas at every zoom level. You can edit node titles inline at any zoom level — press Enter or double-click a node without needing to zoom in. Because each view draws a card at a different height, the canvas learns those heights separately as you browse, a few at a time in the background rather than pausing to measure everything at once. In a long timeline that means spacing keeps getting more accurate the more you scroll and the more views you visit, and switching between views stays quick. When a correction does land, the thought you were reading holds its place on screen — the spacing changes around it rather than sliding it away.
Smart detail
Like a map showing major cities before small towns, important thoughts keep more detail while zoomed out. A thought's importance comes from its connections, backlinks, nested children, how much you've written in it, and how recently you created or edited it — so within the same cluster, the busiest and freshest thoughts bloom into richer views first as you zoom in, while lightly connected thoughts wait for closer zoom. The most important few become landmarks that keep their name visible even when everything else has shrunk to bare dots. Importance changes when a thought shows more detail, never how big it draws — every view keeps its one consistent size. Nothing is ever hidden: every thought stays visible at every zoom. Timeline and Outline modes keep uniform detail, and Gravity has its own physics, so smart detail applies in the other arrange modes. Find it under canvas settings → Canvas → Smart detail; new browsers start with it off. Town view is the exception in the other direction: as a map of your whole hierarchy it always uses smart detail, regardless of the toggle.
Focus dimming
When you select or edit a thought, the canvas gently fades everything except that thought, its directly connected neighbors, and the connections between them — like a map highlighting the place you tapped. Click empty canvas to clear the selection and restore full brightness. Focus dimming applies in every arrange mode except Gravity, which has its own focus behavior. You can turn it off under canvas settings → Canvas → Focus dimming.
Arrange modes
The canvas toolbar or command bar (Cmd+K, type >) lets you switch between layout modes: Manual, Outline, Grid, Force, Tree (vertical and horizontal), Radial, Cluster, Timeline ↓ (vertical journal), Timeline → (horizontal timeline), Gravity, and Town. You can sort nodes by title, type, created date, updated date, or connection count — and when the thoughts on the canvas carry typed properties (like the date on Day, Week, Month, and Year thoughts), each sortable property shows up in the Sort by menu too, so a Daily garden can sort by its actual date field instead of the title. Thoughts without a value for the chosen property always sort to the end. You can also group nodes by type or cluster. These settings are saved per-garden on each device, so each workspace remembers its layout. Your hand-placed arrangement is the one that is shared: it is saved with the garden and synced to every device and collaborator. The automatic layouts (Outline, Grid, Tree, Timeline and the rest) never touch it — they only arrange what you see, so a phone left in Manual keeps its arrangement while a desktop works in Outline, and choosing Layout: Manual anywhere brings back exactly what you left. A thought created while an automatic layout is active gets its Manual spot near the middle of your arrangement, nudged clear of anything already there. If an earlier version overwrote an arrangement on one of your devices, that device may still hold a copy: run > Layout: Recover hand-placed positions saved on this device from the command bar (it can be undone).
Town view
Town view shows your garden's entire nested hierarchy on one living map. Every level of nested thoughts loads onto the canvas and is arranged instantly — children orbit their parent in a compact spiral, bigger branches sit nearer the center, and nothing overlaps. Connected thoughts pull toward each other too: when two branches are linked, they're placed in the same neighborhood of the map instead of on opposite sides. There's no waiting for a layout to settle: the map appears in its final shape right away, even with hundreds of thoughts. The arrangement is remembered — reopening Town view (or reloading) restores it instantly instead of re-arranging. Drag a thought and the map makes room with minimal shuffling: overlapping neighbors step aside while everything else stays put. The map reads like a real one: busier thoughts (more content, links, and connections) reach more detail sooner as you zoom, and top-level thoughts keep their names visible even from the farthest zoom, like city names on a country map. Nesting depth drives detail too: every thought stays visible at every zoom, but deeper thoughts carry less detail until you zoom closer — and thoughts stay at compact detail even fully zoomed in, keeping the map scannable (open the side panel to edit content). Town positions live only in this view — connecting and deleting are disabled while it's active, though selecting and editing thought content work normally. Town view works on phones too; to stay smooth it loads fewer thoughts, and very large hierarchies load their upper levels first and stop at a device-friendly limit, so you always get the top of the map instead of a slow or stuck canvas — zoom or drill into a branch to explore deeper. Switch back to any other arrange mode to return to the normal single-level canvas.
Filtering by type
The funnel button in the canvas toolbar (and Filter in the canvas right-click menu) opens the type filter — the same searchable type list you use when assigning types to a thought, including subtypes. Every type starts checked; uncheck a type to hide those thoughts from the canvas, and use the All and None shortcuts to reset or clear in one click. Thoughts whose type isn't in your Types garden are governed by the "untyped" row. Filtering is a view — nothing is deleted, and unchecking it back restores everything.
Timeline modes
Timeline ↓ stacks dated thoughts vertically like a journal — newest on top, with a right-side date rail showing day, month, and year markers as you zoom. Timeline → arranges them horizontally as a chronological band, oldest to newest left-to-right, with the date rail above the cards. Both modes order thoughts by their date property (thoughts without one keep title order), and you can pick a different sort from the Sort by menu afterward. The side rail follows whatever you sort by: dates show the familiar day, month, and year labels, sorting by title turns the rail into an A–Z index, sorting by type labels each group of types, and other properties label each run of equal values. Entering a vertical timeline for the first time lands at full column width on today's entry — or the latest one if today hasn't been created yet. Once you adjust the zoom yourself, the timeline remembers your choice on this device and comes back to it, like every other garden view. Click the + buttons on the rail between dates to fill in missing days. Days with no entries collapse into a single + so the view stays tight.
Nested canvases (sub-gardens)
Every thought can become its own canvas. Drill into any thought to enter a focused sub-garden where you can add child thoughts, make connections, and organize ideas independently. The breadcrumb bar at the top shows your current depth (e.g. Home > Research > Topic) and lets you navigate back up to any level. On smaller screens, intermediate breadcrumbs collapse behind a '...' menu.
Expand and collapse
Nodes connected by edges can be collapsed to hide their downstream children. Click the chevron arrow on the right side of any parent node to toggle its children. This works at every zoom level. Collapsed nodes smoothly fade out and the layout re-arranges automatically. Dragging a collapsed node carries its hidden children along with it, so in Manual arrangement expanding it again opens them around the node's new spot in the same arrangement as before (automatic layouts re-arrange them on expand as usual). Use Cmd+K search to find a collapsed node — its ancestors expand automatically to reveal it.
| Shortcut | Action |
|---|---|
| Cmd/Ctrl + Left or Up | Collapse selected node |
| Cmd/Ctrl + Right or Down | Expand selected node |
| Cmd/Ctrl + Shift + Left or Up | Collapse all |
| Cmd/Ctrl + Shift + Right or Down | Expand all |
Viewport fitting
Press F to fit all visible content into your viewport. If you have nodes selected, it fits just the selection. The fit accounts for any open sidebar or panel, so content is always centered in the visible area.
Context menus
Right-click (or long-press on mobile) on nodes, edges, or the canvas background to access context-sensitive actions. Every menu starts with AI… (Cmd+K), which opens the assistant with what you clicked pinned as its context — the node, the selection, the edge's two endpoints, or the whole garden for the background. The node menu covers the node's main actions: Edit, Open as Garden, Add Child, Open in Panel, Suggest Ideas (AI), Move Into…, Connect To…, Disconnect…, Duplicate, Select Children, Collapse/Expand Branch, Lock Placement, and Delete. Right-clicking or long-pressing a node that's part of a multi-selection opens the selection menu instead, with Align & Distribute, Tidy Up, Arrange (Outline, Grid, Force-directed, Tree ↓/→, Radial — one-shot layouts of just those nodes, ordered by the Sort by field beneath it and spaced like the matching canvas-wide mode), Sort by, Move Into…, Connect To…, Create Region (wraps the selection in a color region, leaving room at the top for the region’s own handles — stopping short of any thought you did not select, so it never takes a neighbour in with it — Manual layout only), Duplicate, Lock Placement, and Delete. Edge context menus let you flip direction, insert a node on the edge, or delete — reachable by right-click on desktop and by long-pressing the line on touch. The canvas background menu lets you create new thoughts, add a color region (Manual layout only), bookmark the current view, paste, and control zoom, layout, sorting, and filters; a color region has its own small menu at its top-right corner for arranging the thoughts inside it (Suggest Connections moved out of this menu and the bottom toolbar — ask the assistant, or run it from the command bar). AI suggestion chips get their own short menu — Accept, Dismiss and Delete. Accept and Dismiss are the same two decisions as the ✓/✕ on the card, which is useful when zoom has shrunk those to a pill. Dismiss is a verdict on the idea: it remembers the rejection so that angle is never offered again, and grows a replacement in its place. Delete just removes the card — nothing is recorded, so that angle stays in the running and a later tier may offer it again. Color regions leave AI suggestions alone entirely, and the suggestions leave regions alone: an un-accepted suggestion is never laid out by a region, never carried along when you drag one by its grip, and never counted when a region works out how big it has to be — the tier beside its thought is what places it, and accepting it makes it an ordinary thought that regions treat like any other. Select several suggestions and the menu offers only Delete (to reject an angle for good, dismiss it with ✕ — that's what's remembered), and the selection pill drops its layout buttons: a suggestion is positioned by the tier column beside its thought, so aligning, arranging or locking one would just be overwritten.
Move Into…
Move Into… (in the node and selection context menus) moves the selected thought(s) inside another thought — pick any existing thought from the searchable dropdown, or choose '+ Create' to make a new container on the spot. The container stays where your selection was: if it isn't already on this canvas, a reference placement of it appears at the selection's position, with your thoughts nested inside. Deleting that reference later doesn't delete the thought — it just removes it from this canvas, turning the operation into a pure move. The one exception is Home: it isn't a thought, so choosing Home (shown when you're inside a sub-garden) moves the selection out to the home canvas with nothing left behind. The dropdown searches across every garden you can reach, and you can move your own thoughts into a garden someone shared with you — that's contributing. Contributing is a one-way gift, though: a garden's owner owns everything inside it, so once your thought is in their garden it's theirs, and moving it back out (including by undoing the move) is refused with a message saying why. Moving someone else's thought out of their garden and into yours is refused for the same reason.
Connect To…
Connect To… (in both the node and selection context menus) connects a thought — or everything you've selected — to one other thought at once — pick any existing thought from the searchable dropdown, or choose '+ Create' to make a new one on the spot, then choose what your selection is to it: 'As parent' makes the selected thoughts the parents of that thought, 'As child' makes them its children (the wording follows the command — “connect these to X as parent” means these are the parent). If the thought isn't on this canvas yet, a reference placement of it lands just below your selection (selection as parent) or just above it (selection as child), because both ends of a connection have to live on the same canvas. Pairs that are already connected that way are skipped, and a single undo removes every connection the command made (one more undo removes the target it placed or created, if any). In Town view you can still connect the garden's own thoughts; the expanded bubbles from nested gardens aren't connectable, and choosing one tells you to switch layouts first.
Canvas background
Canvas settings → Canvas has three toggles for how the canvas reads: Radial rings (the distance rings around center), Type zones, and Color regions. Type zones glow a soft color behind each Type's most significant thoughts — sized by how much of the garden that type represents, with the heaviest type's glow drawn up to twice as large as the lightest so relative weight reads at a glance — so heavily-used types read as tinted areas of the canvas without any manual setup. Color regions are ones you draw yourself: right-click the canvas and choose "Add Color Region" to place a named, colored zone. Each region can also arrange what it holds: the icon at its top-right corner opens a menu with Outline, Grid, Force-directed, Tree ↓, Tree →, and Radial, plus a Sort by list. Every region wears its name above its top edge and keeps its grip and arrange icon just inside it, whatever arrangement it has — so a name longer than the region is wide can never cover either control. A region keeps a row inside its top edge for those controls, and a region nested inside another is given room above it for its name, so it never lands on the row above. Pick one and the thoughts inside line up, and the region sizes itself to hold them — growing when they need more room and closing back down when they need less, every time it lays itself out. A region you arrange is one you have handed its size to. A Manual region is still yours: it grows to hold what you put in it and never shrinks on its own, because that would quietly leave something outside. "Fit region" is how you resize one to its contents by hand, and on a Manual region it is the only thing that makes the box smaller. It sizes to the cards as they are DRAWN at your current zoom, both ways — fit while zoomed out and the region is left wide enough for the chips you can see, and a chip hanging out of it is pulled in rather than left hanging. The layout sizes each card as it is drawn at your current zoom — the full card, the compact card, the chip, or the dot — and an arranged region re-arranges itself whenever the zoom crosses into a different view, a card opens or closes its own body, or a thought’s connected children are folded away or brought back, so its cards never overlap at the zoom you are looking at and a grid fills the region again when you zoom back in. Those automatic re-arranges are not undo steps; only the changes you make yourself are. An arrangement is worked out on your own screen, for the way your cards are drawn at your zoom, and it is not saved: each device lays the region out for itself, and the region is arranged again whenever you open the canvas. What everyone shares is the region itself — where its top-left corner is, which arrangement it has and which thoughts are in it. That is everything a region needs to draw itself, so your collaborators see the same region holding the same thoughts, sized and laid out for their screen rather than yours. Its SIZE is shared only while it is Manual, because then the size is something you drew rather than something worked out: drag a Manual region’s corner and everyone sees the new box, live. An arranged region works its own box out on each screen, at that screen’s zoom, so it is wide enough for the cards as they are actually drawn there — no chip left hanging over the edge when you are far out, and no room left over when you come back in. Dragging an arranged region’s corner therefore changes it here only, the same as any other part of its arrangement; what you drag the corner OVER still joins it for everyone, because that is you deciding what belongs. Moving a region still moves everything inside it for everyone, because the region itself moved. Put an arranged region back to Manual and the box you are looking at becomes the shared one, so everybody gets the size you can see rather than the one it had before you ever arranged it. There is one thing neither the automatic sizing nor "Fit region" will do, and that is change what a region is nested in or what is nested inside it — if sizing to what is on screen would reach over a neighbouring region’s top-left corner, or would take the region out of the one it sits inside, it goes only as far as the full cards need instead. That is why an arranged region zoomed right out can still leave a chip over its edge: nesting decides which region a thought you drop lands in, and that answer is everyone’s, so it is never allowed to depend on your zoom. A region remembers which thoughts belong to it, and that list is shared with everyone who sees the garden. A thought joins a region when you drop it inside, create it there (a new thought, a placed one, a day card, one the AI adds), draw the region around it, or drag the region’s corner over it — and leaves when you drag it out. Dragging the corner IN no longer lets a thought go, because the corner no longer gets that far: a region stops shrinking at the box that still holds what it owns, so its contents can never be left sitting outside it. That stop happens while you drag, not after you let go, and it goes by the cards as they are DRAWN at your current zoom — the same thing "Fit region" measures — so it does not stop somewhere the chips you can see are hanging out of it. Where the contents are ALREADY hanging out, it goes by the full cards instead, because otherwise there would be nowhere left for the corner to go. A region with a Grid narrows to its widest card, because its columns re-flow as you pull the edge in; it then grows its own height for the rows it ends up with, and widens again if something inside it — a nested region, a locked card — needs more room than a column. Use "Fit region" to close it down to its contents in one step, and drag a thought out when you want it to leave. Nothing automatic changes what belongs: a region growing on its own to hold an arrangement, a change of zoom, or a card folding neither adopts a thought nor lets one go, and a collaborator sees the same members you do even when their zoom draws the region differently. Where a thought counts as inside is judged by its centre; once you are zoomed out far enough that cards have shrunk to chips or dots, it is the thought’s icon that counts, because that is the part you can see and drag. Regions drawn before membership was remembered are converted once, from whatever was inside them, the first time you open that canvas in Manual mode. If a thought ever sits inside a region without belonging to it, dragging it anywhere inside that region makes it a member again — and dragging one around outside a region it still belongs to lets it go. While you drag a card, the region it is about to land in lights up — judged by the card’s centre, or by its icon when it is drawn as a chip or dot, not by where your pointer is — so you can see where it will end up before you let go. Only the card in your hand is shown; cards dragged along with it each land by the same rule. An arranged region also re-arranges by itself when you drag a card into it, out of it, or move one around inside it — it plays the change rather than snapping, unless your system asks for reduced motion. Folding a thought’s children away closes the gap they leave, and unfolding them makes room again — both on your screen only, like every other arrangement, so a collaborator with that branch open still sees it laid out. An arranged region closes down to the smaller layout as it does so; a Manual one keeps the box you drew. It does NOT re-arrange when it grows over a card you never touched — that card is not the region’s and is left where it is until you move one yourself or the zoom crosses into another view. Adding or deleting a card does not re-arrange the region either. Once something does re-arrange it, every unlocked card inside gets laid out, so lock a card to keep it exactly where you put it. A region holding a single thought lays that one out too — it goes to the region’s top-left, where the arrangement says it belongs. One undo takes back the whole thing: the card you dropped goes back to where you picked it up and the region’s other cards to where they were before the drop. The same goes for the re-arrange after a resize, for choosing a new arrangement (undo restores the cards and the previous mode together), and for dragging a region by its grip — the region and its contents come back in one step. "Re-arrange now" runs the same layout again after you have moved things about. Manual is the default and puts nothing in charge of position. Going back to Manual puts the cards that were there when you first left it back where they were; a card you added since stays where it is, because there is nothing to put it back to. Creating a region around a selection holds exactly what you selected — the band it keeps above them for its own name and grip can reach over a thought without taking it in. Regions nest: draw or drag a region so its top-left corner sits inside another region and it becomes that region’s child — nothing to set up, and dragging it out un-nests it. A card inside the inner region belongs to the inner region, so the outer region’s arrangement places the inner region as one box — with room kept above it for the inner region’s name, so it never sits on top of what the outer region laid out in the row above — and never touches the cards inside it; when the outer region moves a nested region, everything inside it moves along, locked cards included — a lock keeps a card in place within its region, and it is the region that moved. When a nested region grows — you drop a card in and it re-arranges, or you resize it — its parent makes room too, all the way up, and one undo takes the whole chain back. An arranged parent makes room by re-arranging; a parent with no arrangement moves nothing, so it grows its own box instead — which is also what happens when you drag a region into it, so the region you dropped is enclosed rather than left hanging over the edge. It does that for a Manual child, whose box everyone agrees on. An ARRANGED child is a different size on every screen, so there is no box to grow to and the parent leaves it alone: a region that arranges itself past the Manual one around it hangs over the edge rather than pushing it out. Give the parent an arrangement too, or fit it by hand, when you want it to follow. Making a region smaller never grows the region around it. Dragging a region by its grip carries the regions nested inside it as well as its cards, and moving a region into or out of an arranged region makes that region re-arrange; while you drag it, the region it will land in lights up, as for a card. Deleting a region hands its cards to the region around it (or to none), and undoing the delete gives them back to the region it brings back; creating a region inside another one lands it there — either way the region around it makes room, by re-arranging if it is arranged and by growing its own box if it is not, each in one undo step with the delete or the add. Under the pointer, the innermost region is the one whose handles you get. Resizing the region by dragging its corner re-arranges it when you let go; a Grid region re-flows its columns live as you drag, so you can see the cards settle before you release. A thought with a locked placement stays where you put it and the region grows to keep holding it, but the rest of the layout does not step around it — so a locked card near where the layout starts can end up underneath one. The arrangement belongs to the region rather than to your browser, so collaborators and your other devices see the mode you chose; anyone with view-only access sees the result but not the menu. Drag the grip handle at the top of the region — it fades in when you hover — to move the region together with every thought inside it, and its bottom-right corner to resize it — the corner grabs from just outside the region as well as just inside, so you can catch it on the way in or out. The name is not a drag handle: it is only a label, so a long one never doubles as a place the region moves from. Double-click the region's name to rename it, and click its color dot to cycle colors. While you're actively zooming, a region's label and handles briefly tuck away, then fade back in once you lift your fingers and the gesture settles (the colored area itself is always visible) — so you can move, resize, and rename regions at any zoom level, including far out where large regions are drawn. Regions only appear in the free (Manual) arrange mode — the auto-arrange modes move thoughts around, so switching to one hides your regions until you return to Manual.