Illustration of AI in government procurement.

Decoding E-Tendering: How AI and Fuzzy Logic Are Revolutionizing Government Procurement

Kai Mendoza monogram
Kai Mendoza in Tech & Innovation October 2026 • 4 min read.

"From Efficiency Boosts to Combating Corruption: Exploring the Cutting-Edge Technology Transforming Public Sector Contracts."


Government e-tendering (GeT) is undergoing a significant transformation, fueled by the integration of advanced technologies. This evolution is critical for enhancing traditional government systems, making them more efficient and accountable. As digital platforms become integral to public services, the need for secure, transparent, and efficient procurement processes has never been greater.

This paper delves into a groundbreaking approach that combines Genetic Algorithms (GA) and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), enhanced with Intuitionistic Fuzzy Information. This hybrid model is designed to optimize the selection of tenderers in GeT systems. The use of Fuzzy Logic allows for the nuanced handling of tenderer attributes, leading to more accurate and fair evaluations.

The significance of this research lies in its potential to enhance the efficiency and fairness of public procurement. By leveraging AI and advanced analytical techniques, governments can make better decisions, reduce costs, and increase transparency. This article provides a comprehensive understanding of the techniques and methods employed.

AI Search Multiple angles on this topic

Procurement AI by the Numbers

Gitnux reports that 58% of US federal agencies use AI e-sourcing. The same source says AI invoice matching has delivered $20 billion in savings while reducing purchase-order processing time from five days to 45 minutes. Procurement Tactics, WorldMetrics, and zeiv.ai likewise frame 2026 as a period of expanding AI adoption, market growth, agentic procurement, and increased attention to procurement readiness and sustainable returns.

Where AI-Assisted Tendering Falls Short

Current guidance presents AI as useful for producing an initial tender draft, but not as a replacement for human judgment. My Business reports that AI-generated drafts can be quick yet generic, lacking the tailored and client-centric strategy needed to distinguish a proposal. Other guidance emphasizes that AI-assisted tender writing requires clear boundaries and mitigation measures, while SafeSourcing highlights data privacy and security risks, including the potential financial and reputational damage from unauthorized access or breaches.

From Dartmouth to Generative AI

The history of artificial intelligence extends from earlier myths and stories about artificial beings to the formation of AI as a research field. The Dartmouth workshop in 1956 is widely identified as a foundational event for the modern discipline. Later timelines trace major developments through the Turing Test, machine learning, deep learning, the Transformer paper, ChatGPT, and generative AI, with some contemporary timelines extending to GPT-6.1 Sol in September 2026.

Unpacking the Power of GA and TOPSIS in E-Tendering

Illustration of AI in government procurement.

The core of this innovative system is the integration of Genetic Algorithms (GA) and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS). GA is used to automatically determine the optimal weights of the evaluation criteria for tenderers. TOPSIS then employs these weights to identify the best tenderer. This ensures that the selection process is data-driven and objective. Both of these techniques use a system called Fuzzy logic.

Fuzzy logic, at the heart of the system, addresses the uncertainty inherent in evaluating tenderer attributes. Unlike traditional methods, fuzzy logic uses 'fuzzy sets' to represent criteria, allowing for a more nuanced and realistic evaluation. Attributes are expressed as Fuzzy Number Intuitionistic Fuzzy Sets (FNIFSs), which handle the vagueness often present in tenderer data. This allows experts to provide more accurate and detailed evaluations, making the system adaptable.

  • Genetic Algorithms (GA): These algorithms automatically determine the weights for different evaluation criteria, enhancing the objectivity of the assessment.
  • Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS): TOPSIS uses the weights provided by GA to pinpoint the optimal tenderer, ensuring data-driven and transparent decisions.
  • Fuzzy Logic Integration: By using fuzzy sets, the system can handle vague or imprecise data, such as subjective expert opinions, leading to more realistic and effective evaluations.
AI Search Multiple angles on this topic

Research Moves from Promise to Process

Recent research describes AI as having considerable potential for business-process development and public competitive tendering, although adoption is not yet pervasive. A review of internet-based procurement processes presents universal electronic tendering as a way to reduce delays from manual work and promote competitiveness. Other work combines event-log process mining with an LLM-based AI agent to translate tender-performance evidence into decision-oriented narratives, while the Open Contracting Partnership notes emerging uses in drafting documents, generating requests for proposals, and reviewing contracts.

Why Automation Is Not Self-Sufficient

AI-enabled tendering should not be treated as automatically accurate, impartial, or secure. Poor data, opaque recommendations, weak integration, and insufficient human review can undermine the efficiency and fairness that automation is intended to improve. Its practical value therefore depends on governance, verification, and clear accountability rather than on technical deployment alone.

Comparing Digital and AI-Enhanced Tendering

Conventional e-tendering primarily digitizes submission, communication, and contract-management workflows, whereas AI-enhanced systems aim to interpret documents and support evaluation. The distinction is one of capability rather than a complete replacement: AI can accelerate repetitive analysis, but public procurement still requires transparent criteria, reviewable evidence, and accountable decisions. Outcomes will vary according to data quality, system integration, and the level of human oversight.

The practical application of this system involves several key steps: First, a decision matrix is constructed, incorporating tenderer attributes. This matrix is then weighted using the weights obtained from the GA. Next, the system identifies both a positive ideal solution (PIT), representing the best possible outcome, and a negative ideal solution (NIT), representing the worst. The closeness of each tenderer to the PIT and their distance from the NIT are then calculated to determine the most suitable choice. The system ranks each tenderer according to their closeness coefficient, and the tenderer with the highest score is selected.

The Future of Public Procurement: Smarter, Faster, and Fairer

The integration of AI and advanced analytical methods into government e-tendering presents a promising future for public sector procurement. The move towards more efficient, transparent, and data-driven processes not only streamlines operations but also reduces costs and promotes fairness. As these technologies continue to evolve, GeT systems are poised to become even more sophisticated, contributing to better governance and more effective public services.

AI Search Multiple angles on this topic

From Document Search to Decision Support

Commercial tender-analysis platforms increasingly position AI as a decision-support layer for procurement teams. TenderAI describes structured scoring, compliance verification, and risk insights with evidence mapped back to the source document, while TenderScorer advertises criterion-by-criterion analysis, indicative scores, and prioritized recommendations in under two minutes. Tender.bid similarly emphasizes faster bid qualification, earlier requirement extraction, and reduced repetitive document review, although these vendor claims describe product capabilities rather than independently verified sector-wide results.

Connected, Automated Procurement

Future e-tendering trends include AI that extracts key data from bid documents, detects abnormal pricing, and improves evaluation efficiency; etender1 reports an improvement of over 50% for this area. The same outlook emphasizes ecosystem integration across internal systems and cross-border services throughout the supply-chain lifecycle. Other forecasts point to stronger transparency and regulatory standards, as well as the integration of AI with blockchain and big-data analytics to support more efficient and secure procurement.

The Infrastructure Problem

The broader challenge is not only whether AI can analyze a tender, but whether procurement organizations can support reliable digital systems around it. Research on e-tendering identifies implementation barriers in construction and links effective electronic procurement with transparency, efficiency, competitiveness, and sustainable development. The supplied sources also identify interoperability with existing systems, data quality, and cybersecurity as systemic concerns that must be addressed alongside AI and fuzzy-logic capabilities.

Augmentation, Not Abdication

A case study of tender-evaluation AI reports measurable improvements in transparency, efficiency, and evaluation processes when generative AI is used with appropriate safeguards and regulatory alignment. Reported procurement use cases include supplier-performance and compliance monitoring, market-risk identification, spend classification, invoice automation, contract review, spend analysis, RFP automation, and supplier-risk assessment. Tender360.ai describes case studies involving faster tender creation, improved supplier evaluations, accelerated proposal responses, and reduced operational disorder, but these examples remain product- or study-specific rather than proof of uniform results across government procurement.

What the procurement evidence says about the method

A real-life scenario from Serbia is presented to demonstrate the proposed method’s feasibility in public procurement. That makes the method relevant to an actual procurement setting, but the cited fact does not report how well it performed or whether it outperformed another approach.[1]

The choice of evaluation criteria and their weights can shape who wins: a committee using MEAT may favor a bidder by assigning high weight to a criterion that bidder alone can fully satisfy. This describes a risk in the approach, not evidence that such favoritism occurred in the Serbia scenario.[1]

How the Serbia example relates to procurement guidance

The Serbia scenario gives the method an applied public-procurement setting, while the MEAT discussion identifies a way criterion weighting can create bidder preference. Together, they distinguish demonstrating use in context from establishing that the weighting process cannot be manipulated.[1]

OECD material describes machine learning as a way to evaluate supplier performance by finding patterns in procurement data, including delivery delays, quality issues, and cost deviations. That is a supplier-monitoring use, distinct from selecting among tenderers, and the cited description does not establish its performance in the Serbia example.[2]

What evidence could clarify the method's procurement risks

  • How would data flaws or bias be detected and managed when automated tools analyze supplier records? The guidance calls for addressing or planning around data issues, while the described analysis relies on procurement data to identify performance patterns.[2] [3]

Sources

  1. Springer (acta.uni-obuda.hu)
  2. Digital transformation of public procurement (EN) (www.oecd.org)
  3. Guidelines for AI procurement - GOV.UKGOV.UK (www.gov.uk)

About this Article -

Written with AI assistance from published research, and reviewed by the Mystum team. See our About page for more information.

Everything You Need To Know

1

How is the integration of Artificial Intelligence and Fuzzy Logic improving government e-tendering (GeT)?

The integration of Artificial Intelligence and Fuzzy Logic is revolutionizing government e-tendering by making it more efficient, transparent, and secure. Specifically, techniques such as Genetic Algorithms (GA) are used to optimize the weighting of evaluation criteria, while Fuzzy Logic allows for a more nuanced handling of tenderer attributes, leading to more accurate and fairer evaluations. This helps reduce costs, improve decision-making, and enhance the overall integrity of public procurement.

2

What role do Genetic Algorithms (GA) and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) play in optimizing the selection of tenderers?

Genetic Algorithms (GA) are used to automatically determine the optimal weights of the evaluation criteria for tenderers, making the selection process more data-driven and objective. Then, Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) employs these weights to identify the best tenderer, ensuring a transparent and fair decision-making process. This hybrid model enhances the efficiency and fairness of public procurement by leveraging AI and advanced analytical techniques.

3

How does Fuzzy Logic address the uncertainty inherent in evaluating tenderer attributes?

Fuzzy Logic addresses uncertainty by using 'fuzzy sets' to represent evaluation criteria. Unlike traditional methods, Fuzzy Logic allows for a more nuanced and realistic evaluation of tenderer attributes. These attributes are expressed as Fuzzy Number Intuitionistic Fuzzy Sets (FNIFSs), which can handle the vagueness often present in tenderer data. This enables experts to provide more accurate and detailed evaluations, making the system more adaptable and effective.

4

Can you explain the practical steps involved in applying the integrated system of Genetic Algorithms (GA) and Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) with Fuzzy Logic?

The practical application involves several key steps. First, a decision matrix is constructed, incorporating tenderer attributes. This matrix is then weighted using the weights obtained from the Genetic Algorithms (GA). Next, the system identifies both a positive ideal solution (PIT), representing the best possible outcome, and a negative ideal solution (NIT), representing the worst. The closeness of each tenderer to the PIT and their distance from the NIT are then calculated to determine the most suitable choice. The system ranks each tenderer according to their closeness coefficient, and the tenderer with the highest score is selected, ensuring a data-driven and objective selection process.

5

What are the potential future implications of integrating AI and advanced analytical methods, like Genetic Algorithms (GA), Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), and Fuzzy Logic, into government e-tendering?

The integration of AI and advanced analytical methods presents a promising future for public sector procurement. By moving towards more efficient, transparent, and data-driven processes, operations are streamlined, costs are reduced, and fairness is promoted. As technologies like Genetic Algorithms (GA), Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS), and Fuzzy Logic continue to evolve, government e-tendering systems are poised to become even more sophisticated, contributing to better governance and more effective public services. This will lead to smarter, faster, and fairer procurement processes, ultimately improving public trust and resource allocation.

Newsletter Subscribe

Subscribe to get the latest articles and insights directly in your inbox.