Glossary
Plain-language definitions of every term you will see in the Document Blueprint documentation.
Last updated
Plain-language definitions of every term in the Document Blueprint documentation.
Core vocabulary
Activity timeline — The chronological per-case feed of notes, file events, generations, form submissions, and collaborator changes. Composed at read time from several Firestore sources (case notes, file store, template runs, override markers, the case activity subcollection) — not a single canonical log.
Automation — A binding of (trigger) → (mailbox + category + template + source + duplicate-handling). Gmail and Drive automations push files into cases without manual upload. Both categoryId and templateId must resolve at ingestion time or the file is skipped (fail-closed). See What automations are.
Autofill — The engine that fills templates with a case's data and produces an output file. Invoked from the review modal (per file, per case, or in batch from the dashboard) and from Quick Fill in the template editor sidebar for one-off runs without a case.
Batch field — A repeating-row field on a template. Used when the output has a variable number of rows like line items. The latest extraction replaces the entire row array — there is no per-row merging across files.
Capability — A boolean flag on a template: hasExtract (it pulls AI-extracted data from incoming files) or hasGenerate (it has at least one source tab producing output PDFs). A template can have either, both, or neither. Fill (collecting answers via source='question' fields) is implicit in the fields themselves and is not a tracked capability.
Case — A unit of work. Holds files, extracted data, notes, generated outputs, team assignments, and client shares for one thing.
Case identifier — The single top-level value (case.identifier) that uniquely identifies a case within a workspace. Backed by an atomic Firestore lock so two files arriving with the same identifier can never produce two cases. See How cases merge.
Category — A routing label on files. Decides which templates can process the file, which collaborators can see it on a case, which automations target it, and which files a send template attaches. The single connecting tissue across five subsystems.
Clients panel — The case detail section listing external collaborators with portal access. Each row expands to manage Forms to fill — which form templates that specific client can fill from the portal.
Collaborator — Someone with scoped access to specific cases. Workers (team members) appear on the Team panel; external people appear on the Clients panel. Visibility is scoped by file category.
Computed output — An output (on any field) whose value is derived from upstream outputs via an expression — evaluated at extraction time, plus re-evaluated live whenever the expression references $today or $now. Configured as a computed-kind output in the field's Outputs section.
Source tabs (formerly: document combiner) — The ordered list of sources that compose a template's output, shown in the document bar directly above the template editor's preview canvas. The bar picks one of three layouts automatically: multi-source (full tab strip), single-source (compact row with a ⋯ menu, used when there's exactly one source and no form), or form-only (minimal label, used when the template is form-only). Each tab references either a workspace document source asset, another template's output (for aggregating templates), or a blank-page placeholder. A Form tab appears at the leftmost position when the template has form-question fields or non-empty AI Instructions. Dangling references render blank by design.
Document source — A workspace-level PDF asset that templates reference for their base pages and that automations target for ingestion. Stored at users/{uid}/document_sources/, scoped to a workspace via the workspaceId field. Manage in the workspace's document sources panel.
Extract — A template capability. Pulls AI-extracted structured data from an incoming file into the case. Drives the Extract section of the template settings drawer.
Field — A named concept on a template. Has a type and a source (question / extracted / static), plus one or more Outputs that produce the actual values written to case.data. Renaming a field does not migrate existing case data — the data path stays bound to each output's stable key.
Field origin — Provenance metadata on every extracted or user-modified case value: from_extraction (AI set it) or modified (a human edited it). Re-extraction overwrites from_extraction rows but skips modified rows (the case.data._overrides map, keyed by output key, records the skip-list).
Field source — Where the field's primary value comes from. Three values: question (the user fills via a form), extracted (the AI pulls from a document or other context), static (a fixed value baked into the template). Replaces the legacy 8-value valueSource enum that existed before the 2026-05-17 outputs refactor.
Output — A value-producing rule on a field. Each output gets its own stable key in case.data, its own value type, and its own kind — see Output kinds below.
Output kind — How an output produces its value. Four derivation kinds: static (output.staticValue), prompt (derived from an AI prompt over upstream outputs), computed (derived from an expression), binding (read from non-PDF context: email metadata, email body, system clock, or company profile). Primary values from the field's inputs[] (user form input, bulk-extracted value, or per-input static) are stored under each input's key and don't go through the outputs[] list.
Filter preset — A saved filter over a workspace's case list. Each preset is a named set of filter clauses (e.g. status=review AND due_date < today). Distinct from a mailbox, which is a physical partition rather than a saved filter.
Forms to fill — The portal section where clients complete form templates a worker has exposed to them. Submissions write back to the case via the field IDs on the form and land on the activity timeline.
Generate — A template capability and a button. Assembles a template's source tabs against a case, producing a PDF output and recording a pdf_generation activity event.
Identifier lock — The Firestore document at users/{uid}/identifier_locks/{workspaceId}__{identifierValue} that guarantees uniqueness of a case identifier within a workspace. Atomically claimed at case creation; atomically released and re-claimed when an identifier is renamed.
Inbox — The default inbox mailbox in every workspace. Holds incoming cases until they're routed to a custom mailbox or moved to Archive / Trash. Trash is auto-purged after 30 days.
Mailbox — A persistent partition of a workspace's cases. Four kinds exist: inbox (the default landing), archive, trash, and custom (workspace-defined). The sidebar lists all mailboxes plus a synthetic "All Mailboxes" entry. Cases are routed to mailboxes by automations or by user drag-and-drop.
Organization — The tenant layer that holds members, roles, capabilities, billing, and portal association. Mostly invisible — surfaced through Settings → Account → Team and Billing.
Override — A worker edit that locks an output value against future re-extraction. Tracked in case.data._overrides[outputKey]. Clear an override (via the field's history popover) to let the next extraction re-overwrite the output.
Portal — The client-facing surface at /portal. External collaborators sign in here to see cases shared with them. Same email can be a Worker in one organization and a Client in another.
Prefill policy — A setting on a field that controls what autofill writes at fill time: always_prefill (write the resolved value), prefill_if_value (skip if empty), or leave_blank_for_client (skip entirely; reserved for client-facing forms).
Quick Fill — A one-off autofill run that takes a template + one file and returns a filled PDF without creating a case. Lives in the template editor's sidebar, not in the dashboard. Sometimes referred to internally as "Lite Autofill" — same feature.
Re-extract — The button on the case's Extraction Runs panel that re-runs AI extraction on a specific file. Honors the _overrides map — fields you've manually edited are skipped.
Routed from — Read-only chips on a template's settings drawer showing which incoming file categories flow into this template via its source tabs.
Send — Composing a {to, cc, subject, body, attachments} payload from a case and creating a real Gmail draft in the signed-in user's own Gmail mailbox via the Gmail Drafts API. Gmail's compose view opens in a new tab with everything pre-filled; the user reviews and clicks Send from inside Gmail themselves. Document Blueprint never sends mail directly. Permission gated per-worker by the Compose Emails capability (Settings → Team). UI labels: "Send" / "Compose Draft" / "Compose share" / "Open Gmail Draft".
Send template — A saved share preset for a workspace: a name, a tokenized subject, a default body, and source rules of the shape {kind, categoryId?, mode} declaring which file categories attach vs. just link. Does NOT carry visible-category lists or exposed-form lists — those are per-collaborator on the case's Clients panel.
Team panel — The case detail section listing workspace members assigned to the case, scoped by file categories.
Template — The blueprint for a generated document. Defines fields, sources, placements, output naming, and capabilities (Extract, Generate). Top-level sibling to workspaces — a template can be linked to a workspace via template.workspaceId, or stand on its own.
Variant — Deprecated term (pre-2026-05-17). Replaced by Output. The concept is the same — additional values a single field produces beyond its primary one — but every derived value is now just another output in the field's outputs[] list, with kind (computed / prompt / binding) replacing the old "variants" subsection.
Visible categories (visibleCategoryIds) — The list of file categories a collaborator can see on a specific case. Drives what they see in files, the activity timeline, inbox chips, and AI chat tool results. Stored on the collaborator record, not on the send template.
Workspace — The top-level container for one document process. Owns cases, mailboxes, columns, filters, send templates, and view configuration. Lives inside an organization. Workspace operations (create / save / delete) require a Pro subscription.