Tender Lifecycle AI Platform

1. Executive Summary

The platform now supports the full tender lifecycle as a single, connected workflow. A procurement officer can create and publish a tender with AI assistance, bidders can review it, verify their identity, and submit their applications and supporting documents, uploaded documents are read and structured automatically, and bids are evaluated through a deterministic engine that produces an auditable ranking with evidence for every decision.

All core modules are built and have been verified through a complete, live end-to-end run — a real tender was created, multiple bidders applied, their documents were processed, and the evaluation produced a final award recommendation with full traceability. Since the last report, the pre-bid query workflow has been completed and the bidder experience has been substantially deepened — a personal "My Tenders" workspace, an officer ↔ bidder clarification loop, the ability to revise or re-apply a bid before the deadline, stage-gated outcome visibility, digital-signature identity verification, public/private & invited tenders, and a company/KYC profile with administrator approval. A new tender import tool lets officers back-test the evaluation engine on historic tenders. The remaining work consists of incremental enhancements to evaluation depth.

2. What the Platform Does

The platform covers each stage of a procurement, with clear separation between the work that people do, the work the AI assists with, and the decisions that are made by deterministic rules so they remain reproducible and auditable.

• Create the tenderAn AI assistant interviews the officer, then drafts the tender document — scope, clauses, eligibility criteria, document checklist, annexure and BOQ formats, and the evaluation scheme — with the officer approving every section.
• Publish and amendThe tender is published to a public bidder portal. Any change is issued as a formal corrigendum, with full version history preserved.
• Bidders applyBidders review the published tender, prepare their submission through a guided apply wizard, and upload their certificates, financial documents and BOQ.
• Documents are readEvery uploaded document is automatically read, structured into data, and indexed — with the page and source recorded for each fact extracted.
• Bids are evaluatedA deterministic engine checks eligibility and compliance, applies the approved scoring scheme across pre-qualification, technical and commercial stages, and ranks the bidders.
• Award with an audit trailThe result includes the reason for every pass, rejection or benefit, backed by the supporting evidence — and can be reproduced without re-running the AI.

3. Delivery Status by Module

4. Key Capabilities Delivered

Tender creation and document generation

• AI assistant that asks the right follow-up questions and drafts each section for officer approval.
• Automatic generation of eligibility criteria, document checklist, annexures and BOQ formats.
• An evaluation scheme defined per tender — Lowest Price (L1), Highest Quote (H1) or weighted Quality-and-Cost (QCBS) — with tie-breaking rules.
• EMD calculated automatically from procurement slabs, with placeholders that the officer fills before publishing.
• Downloadable tender PDF and a formal corrigendum process with full version history.

Bidder participation

• Public portal where bidders read the complete tender document.
• Guided apply wizard with a two-cover (technical / financial) application kit and a readiness check before submission.
• A reusable qualification profile (turnover, experience, address) so bidders need not re-enter details for every tender.
• Eligible categories automatically have their EMD and fee requirements waived at submission.

Bidder workspace, identity and access

• A personal “My Tenders” workspace showing each application's live status, timeline and any action required of the bidder.
• An officer ↔ bidder clarification loop — officers request clarifications and bidders respond in-flow, with outcomes revealed only at the appropriate stage.
• Bidders can revise or re-apply a submitted or withdrawn bid, but only while the submission deadline is still open.
• Digital-signature (DSC) identity verification with optional cryptographic bid signing, and a company/KYC profile approved by an administrator before participation.
• Public and private (invitation-only) tenders with applicant-type restrictions, enforced on listing, viewing and apply.

Document intelligence

• Automatic reading of both digital and scanned documents, including multi-language content.
• Documents are turned into structured, page-cited data with a confidence level for each value.
• Unreadable, protected or corrupt files are flagged for review rather than passed silently.
• Each bidder's documents are strictly isolated — no bidder can see another's submission.

Evaluation and ranking

• Deterministic engine — the AI assists and explains, but the final score, eligibility and ranking are computed by fixed rules, so results are reproducible and auditable.
• Multi-stage evaluation: Pre-Qualification → Technical → Commercial, with commercial bids kept sealed until the technical stage is approved.
• Bidders rejected at an earlier stage are carried forward as ineligible, and ties are routed to the committee.
• Every decision shows its reason and supporting evidence — the source document, page and extract.
• An evidence-required policy ensures self-declared figures are confirmed against documents before a criterion passes.

Exemptions and compliance

• Tender-level EMD and fee waivers for eligible categories (MSE, Startup, MSME and others).
• Per-criterion waiver or relaxation rules, with eligible bidders clearly marked during evaluation.
• Procurement compliance checks built into tender drafting — EMD percentages, performance guarantee, publication windows and purchase preference.

5. Demonstrated End-to-End

The complete lifecycle has been exercised in a live run: a tender was created with the AI wizard and published, multiple bidders applied and uploaded their documents, the documents were read and structured automatically, and the evaluation ran through all three stages to produce a final, fully-explained award recommendation. The outcome for every bidder — pass, rejection, applied benefit and final rank — matched the tender's defined rules and was backed by evidence.

6. Planned Enhancements