AI Platform for grant opportunity analytics at Prigoda

Prigoda, the regional development agency for Primorje-Gorski Kotar County, plays a vital role in identifying and facilitating access to national and EU funding for public institutions. Its team previously relied on time-consuming manual processes to monitor numerous websites and portals for new calls and tenders, which often led to missed deadlines and inefficient dissemination of funding opportunities.

The existing approach involved manually scanning more than 30 websites, extracting data by hand, and drafting summaries for both internal reference and communication with stakeholders. This process was not only inefficient but also prone to human error, creating gaps in funding awareness and reducing the chances of timely application submissions.

The organisation recognized the need to digitalise and streamline the processes to monitor websites and portals for new calls and tenders. Challenges included:

• High manual workload in monitoring over 30 data sources.

• Delayed identification of funding opportunities.

• Lack of summarised, categorised, and searchable data.

• No centralized system for reporting or user communication.

• Growing expectations from public institutions for real-time, actionable funding alerts.

With the support of EDIH Adria, Prigoda participated in a Test Before Invest service focused on accelerating its digital transformation. As part of this service, a prototype of an AI-powered platform was developed to automate the collection, processing and summarisation of funding calls.

The platform leveraged advanced web scraping technologies for data acquisition and integrated large language models (LLMs) to:

• Generate structured summaries of funding calls, including both detailed internal briefs and concise public-facing versions.

• Automatically classify calls based on issuer, application deadline, eligibility criteria, funding amount, and relevant sectors.

• Provide a user-friendly search interface with advanced filtering options.

• Enable weekly reporting, email summaries and newsletter generation.

• Store and manage all outputs within a structured internal knowledge base.

The pilot application demonstrated a robust set of features implemented using Python and Streamlit, with MariaDB for backend data management. The system was designed for secure deployment on local servers or in the cloud and included capabilities such as asynchronous crawling, real-time query filtering, user notifications, and seamless data storage.

The prototype successfully demonstrated the potential of AI to transform Prigoda's operations:

• Time Efficiency: Significantly reduced the time needed to identify relevant funding opportunities, freeing up staff for higher-value tasks.

• Improved accuracy: Automated data parsing minimized human error and improved the reliability of information.

• Centralisation: Consolidated funding calls into a structured, searchable repository, enabling quicker access and better internal coordination.

• Communication: Enabled one-click generation of newsletters and automated stakeholder updates, enhancing outreach and engagement.

• Scalability: Designed with modular architecture, allowing for easy integration of additional data sources and new functionalities.

• Data governance: Recommended deployment on local infrastructure to ensure full compliance with data governance and sovereignty requirements.

The impact went beyond just internal operations. External stakeholders, including municipalities, schools and cultural institutions, now benefit from more timely and tailored information about funding opportunities, thereby increasing their potential for successful applications.

Following the success of the pilot, Prigoda is now well-positioned to transition the solution into full-scale production, with refined specifications and system integrations tailored to its operational ecosystem.