Items & linking

With a schema in place, this is where a collection actually fills up: creating files, adding rows, connecting them into a graph, and uploading the files and images that back them.

Create a file and add rows

1. Sidebar → + File. Name it (e.g. companies), pick the schema type it holds.
2. The grid appears with one column per schema field (in the order you set on the schema page), plus a pinned Relationships column.
3. Click the + Add row row at the bottom. A new row appears immediately, named New <Type> N — click into the name cell and rename it (the name is the item's unique identifier within the file, and what you'll search for when linking to it).
4. Fill in the other cells — they save automatically per-cell, no separate Save step.
5. Click a row's body area (or open it) to write longer notes in the Markdown editor if the type doesn't already have a dedicated richtext field for it.

Link items together

Two ways, same underlying mechanism:

• Per-row: click + link in a row's Relationships cell → search for the target item (across every file in the collection) → pick a relationship type (from the schema's declared RelTypes, or type a custom one) → hard or soft link → optional key/value properties (e.g. role: primary) → Add. The link shows up as a chip on both the row you started from and (via the graph) the target.
• Bulk, from the graph: select multiple nodes in the Surface or Explore graph view (shift-click or drag-select) → the multi-select toolbar's link action → pick one target + rel type → links every selected node to it in one action. Useful for "link these 12 contacts to this company."

Links are directional and typed, and can point at items in a different file, or a different collection entirely.

Files, images, and enrichment

• Any image/pdf field: click to upload. Width/Height (image) or Summary (pdf) appear instantly.
• If the schema field has "automated fields" enabled, a Description and Tags (as links to a Classification type) get filled in asynchronously by a separate enrichment worker — the cell is blank right after upload and fills in on its own once the job completes; no manual refresh needed. Check pending jobs via the sidebar's Queue button.