Overview

Document Processing & AI Extraction is the fourth stage (S04) of the BidFactory pipeline. After creating a bid from a discovered tender, your next step is to upload the original tender documents and let BidFactory's AI engine extract structured data from them.

This module covers three connected capabilities:

• Document Management — Upload, organize, and manage tender documents (PDF, DOCX, DOC)
• AI Extraction — Automatically extract metadata, requirements, evaluation criteria, team needs, pricing formats, and more from uploaded documents
• RFI Auto-Fill — Generate draft responses to Request for Information questions using AI

End-to-End Workflow

Each step builds on the previous one. Documents feed extraction, extraction results inform your bid strategy, and RFI auto-fill uses both documents and extraction results to draft intelligent responses.

Documents Tab

Every bid has a Documents tab accessible from the bid detail page. This is where you manage all tender-related files for the bid.

Navigating to Documents

1. Open a bid from the Bids list (or navigate directly via /bids/{bid-id}/documents)
2. Click the Documents tab in the bid detail navigation
3. If no documents have been uploaded yet, you'll see an empty state with an upload prompt

1 Documents tab — Empty state before any files are uploaded

The empty state clearly shows the upload area and lists the accepted file formats. Once documents are uploaded, they appear in a table with checkboxes for batch operations.

Uploading Files

Upload Methods

BidFactory supports two ways to upload documents:

• Drag & Drop — Drag files directly onto the upload zone
• Click to Browse — Click the upload area to open your file picker

Accepted Formats & Limits

After Upload

Once files are uploaded, the Documents tab shows them in a table with:

• Checkbox — For selecting documents for extraction
• Filename — The original file name
• Type — File extension / MIME type
• Size — File size in human-readable format
• Uploaded — Timestamp showing when the file was added

2 Documents listed after upload — each row has a checkbox for selection

Running AI Extraction

AI Extraction is the core intelligence feature of S04. It reads your uploaded tender documents and automatically extracts structured data across nine categories.

Triggering Extraction

1. In the Documents tab, select one or more documents using the checkboxes
2. The "Extract with AI" button becomes active when at least one document is selected
3. Click the button to start extraction

3 Documents selected — the "Extract with AI" action is ready

Extraction Processing

After triggering, the extraction runs as a background task via the ARQ worker. During processing:

• The status shows as Pending or Processing
• A progress message displays the current step (e.g., "Reading documents...", "Analyzing structure...")
• You can navigate away — the extraction continues in the background
• The Extraction tab updates automatically when complete

Extraction Statuses

Reviewing Extraction Results

Once extraction completes, navigate to the Extraction tab to review results. The data is organized into nine specialized tabs.

4 Extraction page overview — all tabs visible across the top

The Nine Extraction Tabs

Metadata Tab

The Metadata tab shows the high-level tender information extracted by AI, including:

• Tender title and reference number
• Contracting authority (buyer name and address)
• Contract type (services, supplies, works)
• Estimated value and currency
• Submission deadline
• AI model used and extraction cost

5 Completed extraction — Metadata tab with AI-extracted fields

Requirements Tab

The Requirements tab lists all extracted requirements, each categorized and marked as mandatory or optional:

• Mandatory — Must be met for a compliant bid
• Optional — Nice-to-have, may improve scoring

Requirements are grouped by category: Technical, Qualification, Certification, Reference, and Mandatory.

6 Requirements tab — categorized requirements with mandatory/optional flags

Other Extraction Tabs

The remaining tabs provide specialized views:

7 Additional extraction tabs — detailed data across all categories

Editing & Approving Extraction

AI extraction provides an excellent starting point, but you should always review and refine the results. BidFactory supports inline editing for all extracted data.

Editing Metadata & Sections

On any extraction tab, you can edit the AI-generated content:

1. Click on any field value to enter edit mode
2. Make your corrections or additions
3. Click Save to persist changes

All edits are saved via PUT /api/v1/bids/{bid_id}/extract and the reviewed_by field is automatically set to your user ID.

Managing Requirements

The Requirements tab has full CRUD capabilities:

• Add — Create new requirements the AI may have missed
• Edit — Change the text, category, or mandatory flag
• Delete — Remove incorrectly extracted requirements
• Reorder — Adjust the sort order to match document structure

Approving the Extraction

After reviewing and editing, mark the extraction as approved:

1. Ensure all sections have been reviewed
2. Click the Approve button
3. The extraction status changes to Approved
4. Your name and timestamp are recorded as the reviewer

8 Extraction overview with approval state — reviewed and approved

RFI Auto-Fill

The RFI (Request for Information) module helps you draft responses to tender questionnaires. Using the uploaded documents and extraction results, AI generates initial answers for common RFI questions.

Navigating to RFI

1. Open the bid detail page
2. Click the RFI tab in the navigation
3. If no RFI has been generated yet, you'll see the empty state with an auto-fill button

9 RFI page empty state — ready for auto-fill

Triggering Auto-Fill

1. Select the documents to analyze (the same documents used for extraction)
2. Click "Auto-Fill RFI"
3. The AI processes the documents and generates draft responses
4. Processing happens in the background via the ARQ worker

Reviewing RFI Responses

Once RFI auto-fill completes, the page shows the generated responses in a structured format.

Response Layout

Each RFI response includes:

• Question — The original RFI question extracted from the tender
• Answer — The AI-generated draft response
• Confidence — How confident the AI is in the response:
  • High — Strong match with source documents
  • Medium — Partial match, may need manual refinement
  • Low — Limited source data, requires significant manual input
• Source — Which document and section the answer was derived from

10 RFI responses — questions with AI-generated answers and confidence badges

Editing Responses

Review and refine each response:

1. Click on any answer to enter edit mode
2. Improve the language, add specifics, or correct inaccuracies
3. Save your changes

Finalizing the RFI

RFI Statuses

Approving the RFI

1. Review all responses and ensure accuracy
2. Change the status to Reviewed
3. Your user ID and timestamp are recorded

11 RFI approved — status shows "reviewed" with reviewer details

Roles & Permissions

S04 features are governed by BidFactory's role-based access control. The table below shows what each role can do:

Troubleshooting

Common Issues

API Reference Quick Links