Vena’s Game-Changing GenAI Tool for Efficient Data Analysis

Challenge Story

Helping Customers with the Time-Consuming Process of Generating Financial Insights

Vena Solutions, an innovative software company specializing in financial planning and analysis, set its sights on revolutionizing the way financial planning and analysis (FP&A) teams manage their day-to-day work.

Research shows that FP&A teams—Vena’s primary users—spend approximately 75% of their time gathering data and performing administrative tasks. To reduce manual work and streamline operations, Vena was looking to help FP&A professionals automate the intricate process of analyzing and extracting valuable insights from financial and operational data.

To achieve this, they proposed Vena Copilot, a dynamic tool that acts as an extension of an organization’s FP&A team, doing tactical work and producing insights using an intuitive interface to unlock everyone’s strategic potential. They were looking to venture into the cutting-edge world of generative AI – particularly, large language models (LLM) – and sought a partner with the expertise to help fine-tune their suggested solution and deliver within a tight project timeline.

Solution Story

Enhancing LLM Chatbot Precision and Prompt Engineering

Vena Solutions and Adastra began their collaborative journey through Microsoft‘s Unicorn program, dedicated to embracing and supporting startups as they build their business, disrupt new industries, and solve humanity’s most pressing challenges.

After finding a mutual alignment in our cultures, Vena enlisted Adastra, renowned for its leadership in generative AI, to review and fine-tune their solution. Our evaluation focused on the architecture, strategies for efficient generative AI prompt flow and engineering, and refining the precision of the large language model. We also embarked on the process of integrating the sophisticated chatbot within Vena’s existing analytical database technology, CubeFLEX.

Vena Copilot is specifically aimed at enhancing the experiences of professionals in the finance line of business and CFOs. Therefore, the enhancement of LLM precision focused on improving the models’ ability to identify pertinent information from user queries and generating follow-up queries for data retrieval from OLAP (Online Analytical Processing) cubes based on these initial prompts.

Vena and their customers faced notable challenges with the excessive lengths of prompts used for extracting entities and generating queries, which compromised output quality and introduced complexities in prompt augmentation.

To overcome these obstacles, Vena and Adastra employed innovative Semantic Kernel strategies. These strategies enabled the large language models to execute code that significantly reduced the necessary context length. This adjustment not only lowered the costs of chatbot queries but also notably enhanced its performance, driving the project toward its successful completion. Through this collaboration, Vena and Adastra were able to make significant strides in streamlining financial data analysis and insight generation, setting a new benchmark in financial management and planning technology.

Benefits Story

Empowering Organizations with Democratized Access to Strategic Financial Insights

Vena Copilot was successfully launched on time, operating as an integral extension of any FP&A team. With an intuitive interface, it helps FP&A teams efficiently gather data, generate detailed reports, analyze trends, optimize forecasts, and answer complex business queries among other functionalities. For instance, it can provide insights into seasonal revenue patterns by product or identify departments with significant actual-to-budget variances within a given period. This groundbreaking tool democratizes access to strategic insights, enabling users across various organizational levels to plan more effectively and make swift, data-driven decisions.

Adastra’s best practices for large language models (LLMs), significantly enhanced personal identifiable information (PII) protection, data security, and privacy, all while managing to reduce costs. This meticulous approach set new benchmarks for accuracy and efficiency in financial data analysis, and Vena’s customers’ FP&A teams now have more time to focus on driving strategic impact for the business. Other benefits of the solution include:

• Enhanced protection and security of sensitive PII data
• Enabled accelerated, data-driven decision-making
• Seamless integration with Microsoft 365
• Custom AI model training capabilities to fit the needs of Vena clients’ use cases
• 30% reduced costs for chatbot (Azure OpenAI) queries
• Increased operational efficiency and business agility
• Delivered within a tight project timeline

The collaborative efforts between Adastra and Vena Solutions laid a solid groundwork for future advancements in the safety and operational efficiency of generative AI applications within finance. Leveraging Adastra’s deep expertise and strategic employment of emerging technologies, the partnership set a pioneering precedent for broader adoption across the office of finance.

About the Client

Vena is the only Complete Planning platform natively integrated with Microsoft 365. Vena streamlines financial and operational planning, simplifies reporting and analysis processes, and provides advanced analytics and modeling capabilities to help business, finance and operations leaders make agile and more informed business decisions. With Vena, you can leverage the power of Excel and AI-powered insights in a unified, cloud-based platform that enhances productivity, collaboration and insights. Over 1,800 of the world’s leading companies rely on Vena to power their planning.