1

TITLE PAGE

Problem Statement Title :

​

Team Name :

Team Member :

​

​

SmartBid Compliance System for Preventing GeM Tender Rejections

HawkAI

Kavin M , Priya

�PROBLEM STATEMENT

2

Government tender portals like GeM (Government e-Marketplace) follow extremely strict and automated document checks.�But most MSMEs and vendors face tender rejection not because they are ineligible, but because their uploaded documents contain small, avoidable errors.

​

​

​

​

​

​

Tender rejections due to minor document mismatches

​

​

​

​

​

Outdated certificates not detected early

​

​

​

​

​

No Pre-Submission Validation → Vendors Only Discover Errors After Rejection

The Result :

​

• ✔ Eligible MSMEs get unfairly rejected
• ✔ Huge financial losses
• ✔ Wasted time and repeated submissions
• ✔ Procurement officers overloaded with manual checks

Core Issue :

Good companies are losing tenders because their documents don’t match perfectly not because they are wrong.

�PROPOSED SOLUTION

3

GovDoc Genie automatically scans multiple tender documents, detects mismatches, missing fields, and outdated certificates.�It highlights errors with exact evidence and gives a clear compliance decision before submission.�This prevents unfair rejections and ensures vendors submit 100% error-free tender documents.

• AI-powered multi-document analysis (GST, PAN, Udyam, Quotation, Signature)
• Smart mismatch detection (name, address, spelling, numbers)
• Missing field alerts (delivery date, signature, price details)
• OCR support for scanned PDFs
• Compliance score + clear decision (Approve / Reject / Needs Docs)

Key Features of the Solution:

​

TECHNICAL APPROACH

AGENTATHON

4

1. Flask:

Core backend framework that handles API routes, document uploads, and communication with the AI engine.

2. Gemini API (google-generative ai):

Provides intelligent reasoning, field validation, summary generation, and final compliance decisions.

3. Pdf plumber:

Extracts structured text (lines, tables, metadata) from digital PDFs for deterministic rule-based checks.

4. Py MuPDF:

Performs page-by-page parsing, line mapping, and captures exact text snippets for evidence generation.

5. Py tesseract (OCR):

Reads text from scanned or image-only PDFs, enabling full extraction even when text is not selectable.

​

6. pdf2image:

Converts PDF pages into images so OCR can process them; essential for handling government-scanned docs.

​

7. Num py & pandas:

Used for efficient data processing, cleaning, structuring, and consistency verification across documents

​

8. Report lab:

Generates polished PDF compliance reports with summaries and evidence.

PROCESS FLOW

5

1️⃣ Document Upload:

User uploads GST, PAN, Udyam, Quotation, and Signature documents in PDF or image format.

2️⃣ Data Extraction & Structuring:

System extracts key details from documents�and converts them into structured data.

3️⃣ Rule-Based Validation:�data is validated against�GeM tender rules and mandatory formats.

4️⃣ AI Cross-Verification (Gemini):

AI cross-checks documents to detect�hidden mismatches and false rejections.

5️⃣ Final Report & Decision:

System generates compliance score, decision,�and page-line evidence report.

IMPACT

6

❌ High Dependency on Agencies�🔹 No pre-check system → vendors pay commissions.

​

❌ Frequent Tender Rejections�🔹 Manual checks miss small mismatches

.

❌ Slow & Manual Verification�🔹 Document review takes 20–30 minutes.

​

❌ No Evidence or Transparency�🔹 Rejection reasons are unclear.

​

✅ Agency-Free Submission�🔹 Key Feature: Self-Service AI Compliance Check

​

✅ Near-Zero Rejections�🔹 Key Feature: Rule-Based + AI Validation

​

✅ 30 Sec Verification�🔹 Key Feature: OCR + Automated Parsing

​

✅ Transparent & Audit-Ready�🔹 Key Feature: Page–Line Evidence Highlighting

AFTER (With

GovDoc Genie)