Iress, a leading provider of financial services software, has leveraged generative AI to streamline its responses to lengthy Requests for Information (RFIs) and Requests for Proposals (RFPs) used in procurement processes.
Iress Utilizes GenAI for Procurement Response Automation
During the Databricks Data Intelligence Day in Melbourne, Jeff Gibson, Head of Business Intelligence at Iress, discussed the company’s initiative to address a genuine business challenge using GenAI.
Gibson emphasized that Iress aimed to tackle a real-world business problem rather than conducting another theoretical experiment. “We wanted to genuinely solve a business use case and gain insights into implementing and governing AI,” he stated.
The company identified a significant challenge faced by its commercial and product teams related to handling RFIs and RFPs. These requests often involve answering between 50 to 950 questions, requiring information sourced from various specialists and documents across the organization.
The conventional process, which can take three to six weeks, involves multiple teams and incurs substantial costs. “We’re often under pressure to deliver responses to clients promptly, leaving limited time for the sales process,” Gibson explained.
To address this challenge, Iress assembled a cross-functional team comprising members from the commercial, product, business intelligence, data innovation teams, and Databricks. Their goal was to develop a solution within a two-day period.
The result was an application named ‘Bella,’ designed to facilitate the querying of data to generate responses to questions efficiently. While there are other methods to query data, such as using a search engine over a spreadsheet or intranet site, GenAI provided a contextual dimension to the process.
Gibson highlighted that Bella was developed using Databricks, AWS for storage, and Hugging Face to integrate AI models. He mentioned that Databricks’ platform has evolved significantly since the creation of Bella, offering enhanced capabilities for future projects.
Apart from addressing the business challenge, the two-day sprint provided valuable insights for the team, informing Iress’ roadmap for GenAI. “Our cross-functional team gained extensive knowledge in a short period, influencing our internal GenAI roadmap for both internal use cases and product development,” Gibson noted.
The project also underscored the importance of robust data management practices in AI initiatives. Gibson concluded, “Effective data management is crucial for successful AI implementation. This project has not only shaped our GenAI roadmap but also influenced our approach to data management and governance.”
