CoP Blog Competition 2025

Topic – Harnessing AI in Public Procurement: Current Applications and Future Horizons

The Ultimate Shield: Can Artificial Intelligence (AI) Make Public Procurement     CBI-Proof, CVC-Compliant, and CAG-Ready?

Partha Pratim Sengupta

Adviser (Contracts)

West Bengal Mineral Development and Trading Corporation Ltd

(Govt of West Bengal), Kolkata

The Procurement Paradox: Navigating the Minefield

Public Procurement is a classic paradox. On one hand, you have to bring maximum value to the table for every Ruppe you spend. You are accountable for each  Rupee –  it is the taxpayers’ money. On the other hand, your existence is like a fish in a fish bowl – everybody is constantly staring at you. Every decision you take, every order you sign, every vendor you select can be questioned by any citizen of India.

A single missed information, delayed approval, or overlooked clause can cast doubt on an honest decision. Procurement officers handle complex tenders, technical specifications, and strict deadlines, often under huge pressure. They know that any mistake could be scrutinised years later by audits or vigilance agencies.

Former Central Vigilance Commissioner N. Vittal had famously termed the much-feared trinity of India’s statutory watchdogs CBI, CVC, and CAG (the three C’s) as the Shani, Rahu, and Ketu of public procurement. This metaphor aptly captures the anxiety that public procurement officers in India suffer from while performing their critical duties in a fair, just and transparent manner. The stakes are high because public procurement involves public funds, demands trust, and is burdened with legal responsibility.

The Hapless Procurement  Guy

It is AI that can serve as a silent guardian of integrity, ensuring that all the processes are not only fault-free but also ‘CBI-proof’, ‘CVC-compliant’, and ‘CAG-ready’. AI doesn’t replace human judgment; it enhances it, making decisions auditable, traceable, and defensible. It functions both as a preventive and a detection tool, identifying and pointing out discrepancies at the right time and offering public procurement officers actionable alerts at every point in time.

The Indian  Procurement Landscape: Scale and Urgency

Public procurement in India is growing at a rapid pace, not only in volume, but also in complexity. The Government e-Marketplace, GeM, a platform for government procurement, completed transactions worth over ₹ 3 lakh crore in FY 2023-24 alone. However, traditional tender evaluation remains the biggest bottleneck in public procurement. Yes, the front-ends have been more or less digitalised, but back-end processes like tender evaluation  still require officers to download a vast number of techno-commercial and supporting documents submitted by the bidders, check them manually for authenticity and correctness and prepare comparative statements. This is labour-intensive, prone to error, and often slow.

Use of AI in this back-end process is not just a distant vision; it’s already happening. GenAI-powered tender evaluation engine today reduces bid assessment times from weeks to hours, extract and validate bidder data accurately, and automate all types of verification.

Global Impact: The World Bank estimates that improving efficiency and reducing waste in public procurement could save at least US$1 trillion annually worldwide.

Eight Illustrative Procurement Cases: How AI Safeguards Honest Officers

CVC often highlights the need for strong record-keeping. Based on my experiences as Chief General Manager (HOD) at Coal India Ltd, the following scenarios illustrate how AI can address various challenges and protect officers from unintentional errors and extraordinary situations by providing the right information at the right time.

Case 1: The Road Project That Became a Roadblock

An approval for urgent repair work was delayed due to workload pressure and was later labelled “unauthorised” by an audit team.

AI Integrated Logging: Automatically logs emergency approvals with timestamps, linked to weather data and geo-tagged photos.

This justifies the chain of action taken by the officer by showing in a transparent way his due diligence and the need for urgent steps.

Case 2: Bid Evaluation Oversight

A human error caused the reversal of the respective rankings for L1 and L2 bidders, leading to allegations of manipulation and a charge sheet issued to the dealing officer.

Auto-Validation & Anomaly Detection: Real-time automated preparation of comparative Statements and cross-validation of manual entries by an AI-powered engine instantly flags posting errors.

This minimises the risk of human errors and ensures an error-free, transparent audit trail.

Case 3: The Hidden Cartel

Multiple bidders submitted suspiciously identical bids and similar pricing. The Dealing Officer followed normal procedures, unaware of possible rigging and collusion that could unnecessarily implicate him in future.

Forensic Pattern Recognition: Advanced AI analyses bidding patterns and identifies whether the same directors’ or partners ‘are present in multiple bidders’ data with the same IDs. This could point out signs of collusion early in the evaluation process and enable the officer to act proactively, reducing the risk of exposure to systemic fraud.

Case 4: Undetected Conflict of Interest Scenarios

An executive approved a high-value bid of a firm owned by a distant relative without knowing their connection. This was later flagged by the CAG, and motives were imputed.

Conflicts of Interest (COI) Mapping through Automation: AI cross-correlates employee HR data, vendor lists, and family links for potential COI alerts before finalisation.

This protects the officer by uncovering hidden conflicts in time and allow him  timely recusal from the case.

Case 5: The Cyclone Procurement Case

Emergency procurement during a natural disaster bypassed prescribed procedures, later raising concerns about non-adherence to rules and manuals.

AI “Emergency Mode”: Logs time and GPS readings of deliveries with justification notes and sends a notification to all seniors in the hierarchy via SMS/email.

This creates an auditable trail clearly showing that the officer acted in good faith and perhaps overlooked procedures in the greater interest of the organisation .

Case 6: The 200-Page Technical Document  Trap

A material technical deviation hidden somewhere inside lengthy bid documents of about 200 pages was missed by the evaluation committee.

Natural Language Processing (NLP) Parsing : An NLP-powered AI system parses lengthy documents at great speed and instantly highlights deviations, and provides objective evaluation.

This reduces reliance on tardy and slow manual technical scrutiny of techno-commercial bids and protects officers from future allegations of wilful oversight.

Case 7: Field Variation Without Approval

Unexpected discovery of rock formations resulted in extra work, which was approved without a competent signed approval.  CAG flagged this as an unauthorised deviation.

Automation of Tracking & Alerting: AI monitors contractual provisions and sends automated alerts for required approvals, documenting a time-stamped chain of action providing fact-based justification, helping to obtain objective post facto approval.

Case 8: Duplicate Invoice, Honest mistake

During an ERP system migration, a duplicate invoice was processed and paid by genuine oversight. Based on a complaint, an investigation was carried out by CBI, and they treated it as a criminal conspiracy and a fraud.

AI Duplicate Detection: AI identifies duplicate GSTINs and invoice numbers, preventing transactional errors.

AI reduce the risks of unintentional mistakes during system transitions and errors due to high-volume data entry.

AI vs. CVC: The Integrity Check

• Solution: An AI system can design and implement a Probity Scoring matrix. Each procurement decision can be assigned a score based on adherence to CVC and GFR guidelines. A high score proves an officer’s and his team’s commitment to a transparent process, serving as the best defence against false allegations of irregularities.

AI vs. CAG (The Value Check)

• Solution: An AI system can calculate Risk-Adjusted Costing for a bid. In complex purchases like EPC contracts, the tender committee no longer needs to justify selecting a bidder who wasn’t arithmetically L1. AI provides a data-driven, verifiable and objective algorithm, transparently demonstrating that the chosen bidder offers the lowest Total Lifecycle Cost, which includes future operations and maintenance costs. This fulfils the CAG’s requirement for Value for Money.

AI vs. CBI (The Fraud Check)

– Solution: AI leverages its advanced pattern recognition capabilities to identify and unearth fraudulent networks that propel collusive bidding that human observers cannot discover. By continuously monitoring these patterns, AI empowers investigators to unearth deeply hidden evidence.

Quantifiable Impact & Policy Imperative

AI-driven transformation is not a distant goal; it represents a current necessity:

• NITI Aayog data shows that AI systems can significantly increase transparency and reduce evaluation cycle times.

• The National Strategy for AI and IndiaAI missions, backed by over ₹2,000 crores in investments, show the nation’s commitment to ethical, scalable AI applications in governance.

The focus now turns to implementing AI ethically, eliminating algorithmic bias, cleaning data and recognising human oversight.

Conclusion: From Fear to Confidence

Public procurement officers no longer need to worry about unintended mistakes or occasional slippage from prescribed procedures. AI offers a digital shield where officers can log every step while working in good faith and identify fraud early. This change enables officers to work more efficiently and confidently.

This is the future of public procurement: a partnership between human judgment and AI support. When fully integrated, AI will help manage India’s public spending with unmatched transparency, efficiency, and fairness, enabling public procurement officers to focus on creating value rather than worrying about protecting themselves.

Join the Conversation: What’s the biggest roadblock in AI implementation? This topic affects everyone in public procurement. Let’s discuss.

Community Question: The CAG is currently using AI/LLMs for active auditing, giving auditors advanced tools. Which AI tool or capability, such as Predictive Risk Scoring or Automated COI Mapping, do you think a Public Procurement Officer should demand and implement now?

Share your top priority below! Like this post if you believe AI is the future of fearless public service, where integrity is strengthened, and anxiety about constant scrutiny is removed!