Tender Evaluation, From Weeks of Manual Scoring to a Same-Day Comparison

Evaluating vendor tenders manually means re-reading every submitted bid, cross-checking eligibility documents — turnover certificates, experience letters, certifications — against the published criteria, scoring each technical proposal clause by clause against the RFP's own requirements, then parsing commercial bids to rank vendors by price. Doing all three stages by hand, even for a modest number of bidders, takes weeks.

Worse, every tender starts from zero: the same evaluation logic has to be re-applied consistently by whoever happens to be reviewing it, and there's no guarantee two evaluators score the same bid the same way.

The Saransh Document Intelligence Agent ingests each submitted bid — scanned or native PDF/Word — and extracts the fields evaluation actually depends on: eligibility documents, claimed technical specifications, commercial line items. A Saarthi orchestration layer then runs the evaluation itself: validating eligibility against the published criteria, mapping each bidder's technical claims against the RFP's own requirements to produce a comparison and shortfall report, and normalizing commercial bids to determine L1/L2/L3 ranking — with a human approval step before any result is finalized.

Evaluating a tender isn't one task, it's three sequential stages — eligibility, then technical, then financial — where each stage's outcome determines whether the next stage even runs for that bidder. Reproducing the same evaluation logic consistently across every bidder, at every stage, is exactly the kind of multi-step, rule-following, decision-branching work an agent handles and a single prompt can't.

Procurement teams get a structured comparison and shortfall report to review instead of re-scoring every bid by hand, and the same evaluation logic runs consistently whether it's applied to three bidders or thirty.