What is Document Blueprint?
Document Blueprint is a config-driven platform that turns document-heavy processes into deterministic pipelines using templates, cases, and AI extraction.
It's built for teams that process the same documents repeatedly (invoices, permits, forms, contracts) where the real work is configuration: what data goes where, what outputs get generated, who reviews what. Define it once and the engine runs it the same way on every new document.
The model
A workspace is the top-level container for one document process, holding its cases, templates, mailboxes, and file-category routing. Teams run one per process, and the dashboard always shows the active workspace, so switching workspaces changes everything you see. Inside it, templates are the blueprints: each defines the fields, their data sources, where they land on the page, and how the output is named, so autofill produces identical output every run. Cases are the units of work that flow through it, one invoice or permit or contract, each holding files, extracted data, generated outputs, notes, and history. The two connect through extraction: when a file lands on a case, the AI reads it into the matching template's fields, bounded by that schema so results stay predictable, and a human reviews before anything generates. Core concepts has the full definitions.
How it runs
Every workspace runs the same loop, whatever the document. A file arrives by Gmail sync, Drive import, or manual upload; an automation routes it to a category; the matching template extracts its fields; a worker reviews and corrects in the case detail panel; and autofill generates any downstream document from those values.
What changes between one workspace and the next is only the template configuration: the same engine fills an invoice cover sheet, a permit application, or an insurance claim. For a full worked example, the Tutorials build a real bid lifecycle end to end.
Why it's deterministic
One engine runs every workspace: the only difference between a construction-invoice and a legal-intake workspace lives in template configuration. Generation is exact: a template writes to the same coordinates every run, because courts, lenders, and accountants reject documents that drift. The opt-in exception is Document reflows, which lets AI reposition fields at fill time to follow shifting content (see source tabs).
Mailboxes are persistent case partitions (every workspace gets Inbox, Archived, and Trash, plus custom ones), and filter presets are saved clauses (status=review, due_date < today, assignee=me) applied over any mailbox view. The portal shares specific cases with outside people (clients, contractors, plan reviewers), giving them a scoped view of only the fields and files you allow, where they can fill in whatever you left for them.
Troubleshooting
Wrong extracted value. Open the case, fix the field manually, and save. Manual edits are protected: later extractions skip fields you've edited.
Output doesn't match the sample PDF. Values land exactly where you place them, with no auto-layout. Drag the write zone on the canvas to the right spot. If the document isn't laid out identically every time, turn on Document reflows in the template's details dropdown (click the template name in the editor toolbar).
A file showed up in the wrong workspace. Automations route by category, so check the automation's trigger conditions or the file's assigned category.
Template change didn't update existing cases. Edits don't retroactively re-extract. Re-run autofill on an individual case to apply new logic.