Improving service delivery with AI 

Using artificial intelligence (AI) and machine learning (ML) to support Canada’s agricultural sector. 

Background 

During a Town Hall with the Minister of Agriculture and Agri-Food Canada in 2008, farmers and food producers reported that government programs were hard to find. Even when found, it was often difficult to understand who was eligible and how they could access these opportunities. These essential resources needed to be more accessible.  

In 2009, Systemscope began working with Agri-Food Canada (AAFC) on AgPal, a web-based discovery tool for Canada’s agriculture sector. When it launched, AgPal 1.0 helped farmers find federal, provincial and territorial programs and services.  

In 2017, we were asked to expand the scope of the AgPal content to include research, market access information, tools and other related topics. This expansion of the content and discovery logic resulted in AgPal 2.0, powered by AI and ML. 

Purpose 

Our primary objectives in developing AgPal 2.0 were to: 

  • Empower stakeholders: We wanted to help a diverse set of farmers, agricultural producers and agri-food professionals find services, programs, research and other resources to support their businesses. In addition to helping them find this information, we had to make it useful and actionable.  

  • Improve content and usability: Our goal was to elevate the content and usability of the AgPal web-based discovery tool.  

  • Reduce administrative burden: We needed to curate more content while reducing human intervention of harvesting, profiling and categorizing. 

  • Collaborate on a shared vision: We knew that there were partners from multiple jurisdictions that all wanted to help farmers succeed—and we worked to bring them together.  

Problem 

Our journey began by addressing issues within the existing system: 

  • Resource intensive: The existing procedure for populating the web tool required a lot of manual work and was also ineffective.  

  • Content omissions: Key resources were accidentally omitted from AgPal 1.0, leading to incomplete information.  

  • Outdated information: A heavy reliance on third parties led to outdated, inaccurate and missing information. 

Solution 

To overcome these obstacles, we leveraged innovative technologies and implemented the following solutions: 

  • AI and ML integration: We used generative AI and ML to extend and automate content discovery, information management and integration. This was pivotal in the transformation of the system. 

  • Leveraged open data: We integrated the National Agricultural Library Thesaurus (NALT) from the US Department of Agriculture (USDA). Combined with ML techniques, this allowed the tool to harvest and profile content from any publicly accessible website. 

  • Multi-jurisdictional content: AgPal 2.0 includes content from federal, provincial, territorial, municipal and international sources. We were able to aggregate content from many repositories into a unified solution without compromising data sovereignty or integrity.  

  • Improved user engagement: We applied consistent user experience (UX), Usability and Accessibility principles to keep AgPal relevant and focused on the needs of its users. 

  • Curated content: By using AI and ML to gather information, staff have more time to thoroughly verify and validate the content. This curation process ensures the content published in AgPal is accurate and reliable, further increasing the quality and usefulness of the platform.  

Impact  

The results of our efforts were far-reaching:  

  • Database expansion: AgPal’s content increased greatly, growing from approximately 500 items to over 3000 items. This expanded database provides improved access to a broader range of programs and services, supporting growth and success in the agriculture sector. 

  • Streamlined operations: The use of AI led to a remarkable 90% reduction in manual effort from staff across provinces and territories. This not only saved time and resources, but also improved overall operational efficiency. 

  • Exceeding user expectations: Results from user testing far surpassed benchmarks, reinforcing the effectiveness of our approach and the value it brought. Users found AgPal 2.0 provided better results than Google.  

  • The art of the possible: A team from AAFC leveraged the curated content from AgPal 2.0 to create an award-winning plan for AgGPT, a chatbot that uses OpenAI's large language models (LLMs).  

Insights 

  • Investments in curated content and robust models: Incorporating robust models such as NALT create the ideal foundation for organizations to harness the power of generative AI and unlock new dimensions of innovative service delivery and content curation. 

  • The transformative potential of AI and ML: These technologies can have a profound impact on improving service delivery and operational efficiency.  

  • We’re better together: This project is a powerful testament to the impact of innovation, collaboration and a shared vision. There elements helped us achieve remarkable outcomes that are making a real difference in the lives of Canadians. 

By expanding this digital service model to other content sets, we can work together to advance the Government of Canada’s client service agenda. Book a call today.