Getting started with Scribe
Scribe runs inside an organization workspace in the PharmFindr console, one Project at a time. This page is the end-to-end checklist: who needs to do what, in what order, before the first document is processed.
How Scribe is organized
- Organization — your company's workspace. Billing, Scribe enablement, AI models, and the storage connection are all set at this level.
- Project — a unit of work inside the organization (for example, one CDMO relationship or one product). Each Project has its own attached folders, reference documents, routing rules, and runs.
- Workflow — a kind of job Scribe can do, such as Summary Generation or Certificate of Analysis Review. Workflows are enabled per Project.
Scribe is available in organization workspaces only. Match is the only product with a personal plan.
Setup checklist
Work through these in order. Steps 1, 2, and 4 are one-time organization setup; the rest are per Project.
- Attach a card and enable Scribe (organization Owner) — open Manage > Billing, attach a payment card if the organization has none, then click Enable Scribe on the Scribe card.
- Make sure a model is available (organization admin) — routing rules need an AI model to run on. Open Manage > AI Providers to connect your own provider, or confirm PharmFindr-hosted models are available for Scribe.
- Turn on workflows for the Project (organization Owner) — open Manage > Scribe, select the Project, and tick the workflows it should run. See Use cases for what each one does.
- Connect your storage (each person who will browse folders) — connect Dropbox from Connections. See Connect storage.
- Attach a folder to the Project (Project Admin) — in the Project's Connections tab, pick the Dropbox folder Scribe should watch and write to.
- Upload reference documents (Project Editor or Admin) — the templates and specifications Scribe fills in and checks against, in the Project's Scribe Config tab. See Reference documents.
- Create a routing rule (Project Editor or Admin) — tell Scribe which files in which folder to process, with which workflow, model, and reference documents. See Routing rules.
- Drop a file into the input folder — Scribe picks it up automatically, or you can start a run by hand from the Project's Scribe tab. See Runs and review.
Who can do what
Access is permission-based. The default roles map to these capabilities, and your admins can grant the same permissions through custom roles or Groups.
| Task | Permission | Default roles |
|---|---|---|
| Enable Scribe for the organization | billing.manage | Organization Owner |
| Turn workflows on or off for a Project | scribe.attach_workflows | Organization Owner |
| See which workflows are enabled | project.list_workflows | Project Viewer, Editor, Admin |
| Attach or detach folders | project.manage_folders | Project Admin |
| Browse attached folders | document.view | Project Viewer, Editor, Admin |
| Upload reference documents, create or delete routing rules | project.manage_scribe | Project Editor, Admin |
| View runs and their output | agent_run.view_results | Project Viewer, Editor, Admin |
| Trigger or rerun a run, comment, approve or reject | agent_run.trigger | Project Editor, Admin |
If a tab shows a permission message instead of its content, ask a Project Admin to grant the named permission to you directly or through a Group.
Organization membership isn't enough. If you see "You're not a member of this project", ask a Project Admin to add you from the Project's Members tab.
What happens next
Once a routing rule exists, every matching file that lands in its input folder produces a run. Each run writes its result back to the rule's output folder in your own storage, as a Word document with tracked-change suggestions (or an annotated PDF, if your reference document is a PDF). Nothing is accepted for you — see Runs and review.
