Apr 29, 2025

Case Study - Marketplanet

Explore the joint project by Softwise.AI and Marketplanet.

Zamowienia360 – AI Assistant Supporting Public Procurement

Softwise.AI, in collaboration with Marketplanet, developed an intelligent AI assistant designed to support public institutions in creating better public procurement notices.

The tool is based on the analysis of over 100,000 documents sourced from the four largest public procurement data platforms in Poland.

The AI assistant helps users understand procurement criteria, select appropriate conditions, and optimally formulate inquiries, delivering high-quality answers based on real market data.

Key Results

The joint project by Softwise.AI and Marketplanet resulted in the development of an advanced AI tool that significantly enhances the quality and efficiency of preparing public procurement notices.

Thanks to the use of artificial intelligence and a massive database containing over 100,000 documents, public institutions can now make better purchasing decisions faster and more effectively.

Below are the key results of the Zamowienia360 implementation:

  • Over 100,000 source documents in one place – the knowledge base covers data from the main public procurement platforms in Poland.

  • AI support for public entities – users receive immediate answers to questions about criteria, budgets, and procedures.

  • High data quality – the system analyzes only the most relevant documents, such as special procurement conditions and contract templates.

  • Semantic search and filtering – users can refine results by procedure type, CPV codes, or budget.

  • Inquiry evaluation – an additional AI model helps users create better procurement inquiries.

  • Source availability – users can click on the source of the answer and directly access the original document.

About Marketplanet

Marketplanet is Poland’s leading provider of procurement technologies, supporting companies and public institutions for over two decades in optimizing purchasing processes, managing suppliers, and building efficient business relationships.

The company specializes in creating comprehensive B2B and e-procurement solutions covering all stages of the purchasing process – from planning and supplier selection to negotiations, contract execution, and settlement.

Marketplanet’s portfolio includes:

  • Marketplanet OnePlace purchasing platform – an open B2B marketplace connecting suppliers and buyers across industries, enabling transparent and efficient procurement processes.

  • P2P (Purchase-to-Pay) systems – automating and integrating orders, invoices, and payments.

  • Source-to-Contract solutions – supporting companies in need of analysis, tender organization, and contract management.

  • Public Procurement systems – ensuring compliant and efficient tender management for the public sector.

  • Analytical and BI tools – enabling the tracking of procurement performance indicators, supplier evaluations, and budget monitoring.

Marketplanet collaborates with over 1,500 companies and institutions, including some of Poland’s largest enterprises such as KGHM Polska Miedź, Orange Polska, AmRest, PGE Polska Grupa Energetyczna, Grupa Tauron, Santander Bank Polska, and hundreds of public offices and administrative units.

Through the creation of innovative products and services, Marketplanet supports the digitization of procurement in Poland, increases transparency in tender processes, and contributes to the development of professional purchasing relationships in both the public and private sectors.

Key Challenges

The Zamowienia360 project responded to the real needs of public institutions that face the complexity of procurement processes every day.

Creating public procurement notices requires precision, knowledge of regulations, and the ability to navigate through extensive documentation.

The project’s goal was to create a tool that would not only shorten the preparation time for notices but also improve their quality and compliance with current regulations.

Below are the key challenges identified and addressed within the project:

  • Vast amounts of unstructured data
    Public procurement notices are published with numerous attachments – from contract templates to detailed procurement conditions. Extracting valuable information was a major challenge.

  • Low quality of procurement inquiries
    Institutional employees often struggled with formulating precise and complete notices.

  • Need for real-time substantive support
    Users expected immediate assistance without the need to sift through hundreds of pages of documents.

  • Solution scalability
    The system had to perform efficiently with very large data volumes while maintaining speed and answer quality.

Process

The implementation of the Zamowienia360 solution for Marketplanet was a complex project that required not only advanced technology but also a deep understanding of the realities of public procurement in Poland. The key to success was combining expert knowledge, precise data selection, and user experience design enhanced by AI.

The entire process was carried out in several stages – from discovery workshops, through document analysis and structuring, to building the user interface and deploying language models that support the quality of user queries.

1. Discovery Workshops with the Client

The project began with workshops with the Marketplanet team, during which both teams jointly analyzed procurement processes and the needs of public institutions. Through in-depth discussions, it was possible to identify typical user problems and determine which elements of procurement documentation are critical for generating valuable responses.

At an early stage, a quick PoC was developed – a simplified version of the solution based on a narrow sample of manually prepared documents. This allowed for fast validation of the concept and approach.

2. Document Selection

Each procurement procedure includes many attachments – not all of them are relevant. During the analysis, two key document types were identified: Special Terms of Procurement and Contract Templates. Only these documents were included in the further process, which significantly improved the accuracy of answers and the quality of data used by the AI.

3. Building a Vector Knowledge Base

The selected documents were processed, segmented into logical parts, and tagged. The resulting structured database was transformed into a vector knowledge base, enabling fast and accurate semantic search. As a result, the system can understand not just keywords, but the context of user queries.

4. Designing User Interactions

The system was designed to be as intuitive as possible. Users can ask questions in natural language, such as "What are the criteria for security services?" and receive accurate answers generated by AI, pointing to specific document fragments. Each answer includes a link to the source, allowing the user to verify it in the original document.

5. Query Quality Evaluation

The final stage included the development of an additional AI component that evaluates the quality of user queries. This means the system not only provides answers but also educates users on how to formulate more precise questions – ultimately leading to better procurement notices and more efficient purchasing processes.

Solution

In response to the needs of Marketplanet and its public sector clients, Softwise.AI developed an advanced AI-powered solution that combines a powerful knowledge base, precise document processing, and an intuitive user experience.

At the heart of the solution is the AI-powered Public Procurement Assistant – a tool that supports the creation of high-quality procurement notices, based solely on verified data from actual procedures.

By combining NLP technologies, semantic search, and a multimodel AI architecture, the solution helps public institutions work faster, more effectively, and in compliance with regulations.


What is the AI-powered Public Procurement Assistant?

It’s an intelligent assistant based on language models that supports public institutions in analyzing and preparing procurement notices. It leverages over 100,000 documents from the four largest public procurement data sources in Poland, allowing users to ask natural language questions and receive accurate, expert-level answers.

The assistant provides on-demand substantive support – without the need for expert consultation or manual document search.

How Does the AI-powered Public Procurement Assistant Work?

Knowledge Source Integration

The solution integrates information from four different procurement data sources. It extracts information about procurement procedures, along with all relevant attachments and metadata.

Extraction of Quality Data

The system uses AI models to understand the content of each document. Only key documents – Special Terms of Procurement and Contracts – are processed further. Other files, such as generic attachments, are ignored.

Additionally, each announcement is enriched with key metadata, such as the number of submitted bids, type of procedure, CPV codes (categories), procurement budget, and contracting authority.

Advanced Document Analysis

The selected procurement documents are then divided into logical fragments. Each fragment is tagged and categorized accordingly.

Semantic Search

Using natural language processing (NLP) technology, the system understands the context of user questions and matches them to relevant document fragments – even when the wording differs from the original text.

GenAI Answers

Based on the matched search results, the AI generates short and precise answers tailored to the user's query. This way, the user receives a ready-to-use response without the need to sift through the full search results. The answers are based strictly on the verified knowledge base, ensuring much higher quality compared to generic AI tools like ChatGPT.

Filtering Results

Users can refine their searches using filters such as:

  • Type of procedure (e.g., open tender, basic mode)

  • CPV codes (procurement categories)

This makes it easy to narrow results down to the most relevant examples.

Sources and Transparency

Every generated answer includes a link to the original source document – increasing user trust and enabling verification of the information.

Query Evaluation

An additional language model analyzes the queries created by users and suggests improvements. This means the tool not only answers questions but also teaches users how to craft better, more precise procurement notices.

Summary

The collaboration between Softwise.AI and Marketplanet resulted in a modern, scalable AI tool that directly addresses the challenges faced by public institutions in creating procurement notices.

Through deep document analysis, accurate data selection, and natural language processing, the solution:

  • simplifies and accelerates everyday administrative work,

  • increases the quality and transparency of procurement processes,

  • ensures regulatory compliance,

  • and builds users’ competence through guidance and education.

The Zamowienia360 project demonstrates how AI – when rooted in real user needs and supported by a strong knowledge base – can significantly enhance the quality of processes in public administration. It’s a prime example of AI being used not as a novelty, but as a real tool that increases the effectiveness and value of public sector operations.

Optimize Your Processes and Grow Your Business with AI

Schedule a free 30-minute consultation to learn how to leverage your knowledge and data to optimize operations and gain a competitive edge. Let’s find the right solution for your business challenges – together.



Contact us!

By submitting my information, I confirm that I have read and understood the Softwise Privacy Policy.

Contact us!

By submitting my information, I confirm that I have read and understood the Softwise Privacy Policy.

Contact us!

By submitting my information, I confirm that I have read and understood the Softwise Privacy Policy.