1. Access the Client Entity
Navigate to the left-hand side menu.
Select the client entity you want to work on.
💡Tips:
You can enable alphabetical sorting from the entity view to make client entities easier to find. Entities will be listed in alphabetical order (A–Z), with numbered names shown first in ascending order (for example: 1…, 2…, 10…), followed by A–Z.
You can collapse the side menu to enhance your view of the file and extraction review experience by clicking the collapse icon.
2. Identify Documents Needing Attention
Look in the source documents column for files that require your attention.
Documents may arrive here by default, through direct upload, or via SharePoint/Google Drive/Email integration.
💡 Understand Document Complexity
Simple files (single AP Bill, AR Invoice, Direct Expense, or Bank Statement) may flow directly into the Extraction column with data already extracted.
Complex files that require page selection or splitting remain in the Source Documents column.
💡 AI Auto-Split
If a file contains multiple documents, you can use AI Auto-Split to automatically separate pages into individual documents, or split pages manually if you prefer more control.
3. Process Complex Documents
For complex files, you can split them into individual extractions:
Select the pages that belong to a single document.
Click the Create Extractions button.
Assign the correct Extraction Type during creation:
AP Bills
AR Invoices
Bank Statements
Direct Expenses
Repeat until all pages in the file are assigned to extractions.
💡 Why this matters
Choosing the correct extraction type at creation ensures Tofu applies the right extraction logic, validations, and downstream workflow for each document. This is especially important in multi-type extraction setups where different document types are processed and reviewed differently.
Mark Pages as Primary or Supporting
For multi-page documents, you can label pages as:
Primary — pages that contain key totals or summaries
Supporting — attachment or backup pages
This helps keep reviews consistent, especially for long invoices or statements where only certain pages drive extracted totals.
📕 For more details, see Label Document Pages as Primary or Supporting
4. Mark Documents as Done
Once all pages are correctly assigned to extractions, mark the file as done.
Hover over pages to ensure each is linked to an extraction.
💡 Automatic “Done” behavior
If a simple file is fully processed automatically (for example, it produces a single clean extraction), Tofu may mark the source file as Done automatically.
💡 Tip: To keep your list clean, use the Status filter in Source Documents to hide Done files and focus only on items that still need attention.
Managing Duplicated Files
If the same file is uploaded more than once, Tofu may label it as Duplicated.
Tip: Always check the original document’s status and mark duplicates as Done to keep your workspace clean.
📕 Want to learn more? See Handling Duplicated Documents in Tofu
5. Move to Extraction Column
Once the Source Documents column is empty, click the button on the top right to access the Extraction column.
6. Confirm Data Extraction
Select newly created extractions from the original file.
Click the extraction icon to confirm they are ready for processing.
The extraction will follow the rules and workflow associated with the selected extraction type.
💡 You can do data extraction in bulk by selecting multiple extractions.
Need help after repeated extraction attempts?
If you re-run extraction multiple times within a short period, Tofu may show a Request help option so you can quickly ask for help with the exact extraction you’re working on.
Open the extraction you’re reviewing.
Hover over the Extract button and click Request help.
Tofu automatically includes the extraction link. Add a short description of the issue and send.
7. Review Existing Extractions
While new files process, review previously accepted extractions (AP Bills, AR Invoices, Bank Statements).
Use the centering functionality to zoom in on specific data sources.
8. Verify Extractions
Ensure all data in the extraction matches the original document.
Verify to complete the review process.
9. Completion and Next Steps
Verified extractions will disappear from the list due to the filter.
You will automatically be directed to the next extraction for review.
A pop-up confirms that knowledge from the verified extraction is saved in Settings → Knowledge Base, under the selected entry type (such as AP Bills or AR Invoices). Contact-related knowledge is stored under Contacts and automatically scoped to the current entry type.
Note: If a document stays in Pending or shows Limit Reached, your organization may have hit a monthly usage limit. Check Billing, upgrade to resume processing, or wait for the monthly reset.