01
Ingestion
Accept PDFs, scanned images (JPEG, PNG, TIFF), ZIP archives of mixed formats. Automatic file type detection and routing.
ProductThe evaluation engine your officers
The evaluation engine your officers
wished they had 10 years ago.
Tender Intelligence is a document evaluation platform — not a chatbot, not a search engine, not a summary tool. It reads documents the way your officers do (but faster), fills evaluation templates the way your officers do (but without fatigue), and provides evidence the way your officers should (with exact page references for every answer).
Every page. Every format. Every language.
The foundation. Before any evaluation happens, every submitted document is ingested, processed, and structured — regardless of format quality.
02
OCR + Vision AI
Multi-layer extraction. Primary OCR for printed text. Vision AI for handwritten annotations, stamps, seals, and degraded scans. Layout analysis preserves table structures, headers, and document hierarchy.
03
Document Classification
Automatic identification of document type — work order, completion certificate, balance sheet, registration document, PAN card, GST certificate. Maps each document to relevant evaluation criteria.
04
Data Extraction
Structured extraction of key fields — dates, amounts, registration numbers, company names, certification details. Cross-referenced across multiple documents for consistency checks.
05
Indexing
Full-text indexing with page-level granularity. Every extracted data point is tagged with its source document and page number. This is what makes page-level evidence possible.
Evaluation Engine
Template-driven. Evidence-backed. Officer-controlled.
The evaluation engine is where the AI does the work your officers currently do manually — matching criteria to evidence. But unlike a human reading page by page, the AI processes all documents simultaneously, cross-references across submissions, and never gets tired on page 200.
For each criterion, the AI:
- 01Searches all submitted documents for relevant evidence
- 02Extracts the matching data (registration number, past works, turnover)
- 03Evaluates whether the criterion is met
- 04Assigns a confidence score (High / Medium / Low / Not Found)
- 05Records the source page reference
The completed evaluation table mirrors your existing template exactly — same rows, same columns, same format. The only addition is the confidence score and page reference columns.
Colour-coded clarity. One-click verification.
Visual Scan
Every evaluation cell is colour-coded by confidence level — green for high, amber for medium, red for low or not found. Officers scan the table visually: green rows get a quick check, amber rows get attention, red rows get investigation.
HighMediumLowNone
Click to Verify
Click any cell to open the source page. The relevant section is highlighted. Compare the AI's extraction against the original document in seconds.
Doc 3 · Page 14 · Line 27
Full Audit Trail
Override any cell. Add notes. Flag for committee review. Every action is logged with the officer's identity and timestamp. The AI's suggestion and the officer's final decision are both preserved — because in RTI, both matter.
Your format. Your workflow. No translation needed.
Export the completed evaluation in the same Excel format your committee expects. Or generate a PDF summary report with all page references included. The output looks exactly like what your officers would have produced manually — except it was done in 20 minutes instead of 8 hours.
GeMCPPPState PortalsAPI AccessBulk ProcessingScheduled Runs