Spar Nord Fonden is a Danish foundation dedicated to strengthening community bonds by supporting cultural, recreational, and social initiatives that make a tangible impact. With a mission to distribute millions in funding annually to meaningful projects across Denmark, the foundation operates within a highly structured framework guided by six core impact principles. Each year, Spar Nord Fonden receives more than 3,000 detailed applications submitted across four deadlines, ranging from smaller local projects to large-scale national initiatives.
The periodic evaluation process posed a significant challenge for the foundation’s team. Limited resources and high seasonal workloads made it difficult to manage the increasing volume of applications efficiently while maintaining their high standards of quality and compliance. The process required in-depth manual reviews, often leaving staff with little time to focus on the foundation’s broader strategic goals. Spar Nord Fonden needed a scalable solution to streamline evaluations and redistribute their resources toward driving greater impact through their funding efforts.
Recognizing the potential of AI to address these challenges, Spar Nord Fonden sought Trifork’s expertise to identify and implement an efficient solution that could streamline their processes without compromising quality. The goal was to manage an increasing volume of applications smoothly and effectively.
To achieve this, they required:
Trifork began by analyzing Spar Nord Fonden’s existing evaluation process using a “ride-along” approach, mapping out the steps involved in assessing grant applications. This identified key pressure points where AI could significantly reduce workload during peak periods.
The solution focused on automating the evaluation process, leveraging AI to assess applications against Spar Nord Fonden’s six established impact principles. With the foundation’s clearly defined policies and well-documented principles as a base, Trifork used prompt engineering to instruct the AI. This ensured sharp, reliable, and actionable assessments closely aligned with the foundation’s standards.
Development followed an agile methodology, with continuous feedback incorporated to refine and optimize the system. The frontend was built with Next.js, and the backend was developed in Python using the most current OpenAI models running on Azure to ensure data is not used for training public models.
The system was hosted on Azure Container Apps and secured with Azure Entra ID to ensure compliance and data security. A modular design allowed for seamless updates to newer, more cost-efficient AI models, improving performance and reducing costs over time.
Automated PDF processing to extract, summarize, and analyze application content.
A “talk to your data” feature, enabling users to make additional queries into the application, get a proposal for an inauguration speech for the project, or a proposal for a social media post about the project
Evaluation scoring based on the foundation’s six impact principles, complete with reasoning for each score.
“The speed and ease of the process were essential, and the result has lifted a significant workload from our team, allowing us to focus on our raison d’être – supporting impactful initiatives. Trifork delivered an effective solution of great value with minimal effort on our side.”
The solution was successfully designed, developed, and deployed in just a few months, with the first functional version going live in April 2024. Through an agile process, Trifork continuously refined the platform based on user feedback, ensuring it perfectly aligned with Spar Nord Fonden’s specific needs.
Delivered in just 175 hours, the project showcased how a relatively small investment of time could create significant value. By focusing on Spar Nord Fonden’s needs and leveraging their existing evaluation principles, Trifork efficiently created a solution that reduced evaluation time by 25%.
The solution incorporates a “human-in-the-loop” approach, ensuring that AI output is reviewed and refined by Spar Nord Fonden’s team. Approximately 90% of the AI’s output is directly used in the further evaluation process, highlighting the effectiveness and reliability of the system in supporting the foundation’s workflow.
This highlights the effectiveness of an iterative approach in achieving impactful results quickly. Looking ahead, the collaboration with Spar Nord Fonden will continue into 2025, with plans to explore additional AI use cases, further building on the strong foundation of rapid, high-value solutions.
This solution is now a part of Trifork’s AI platform, Corax, which streamlines prompt engineering and makes it easier to implement and refine AI solutions across various use cases. The modular nature of Corax ensures flexibility, allowing Spar Nord Fonden to adapt the system to future needs without extensive redevelopment.
Contact
If you prefer, you can also contact us on info@trifork.com
Subscribe
Find out the latest news first
Content