Demystifying AI in Finance: Practical Applications for Finance Teams

With invaluable contributions from Ashutosh Shah & guidance from Ayon Banerjee

The AI revolution has taken the business world by storm, fueled by advancements like ChatGPT that have captivated professionals across industries. Amidst the buzz, finance teams are curious about harnessing AI to enhance their day-to-day operations and drive efficiency. In fact, companies like Sturppy are already piloting an AI CFO. According to a recent McKinsey report, Generative AI's productivity impact could contribute an astounding $2.6 trillion annually.

Here’s my attempt to simplify the AI landscape for finance professionals and provide a practical primer to kick-start your AI journey.

1. Uncover Hidden Patterns in Data: AI empowers finance professionals to identify elusive patterns within their data. Imagine discovering the impact of interest rate hikes on pharmaceutical sales or logistics spending, or understanding how training investments affect branch revenue in banks. AI can unveil these connections, bringing valuable insights to the forefront. For example, an FMCG selling in Chinese markets realized the need to keep the Zodiac beliefs in mind for their revenue forecasting. In certain Chinese markets, more kids are born in the auspicious years and therefore revenue suddenly leaps with a lag of approximately 2-3 years for child nutrition products. AI can help finance uncover similar hidden patterns in data.

2. Identify & Standardize New Financial Ratios: By analyzing data to identify hidden patterns, AI can suggest novel financial ratios for your team to track. With a simple chat interface, finance professionals can calculate these derived ratios on the fly and explore deeper insights. For instance, you can examine branch revenue to training cost ratios in banks, calculate segment revenue to employee satisfaction scores, revenue-to-diversity scores, or lag effects of marketing cuts on revenue. Natural language interfaces enable effortless analysis without the need for complex reports or IT support.

3. Hypothesis Testing and Predictive Analytics: In today's rapidly evolving business landscape, finance teams must constantly analyze multiple scenarios. AI can generate these scenarios swiftly through conversational interfaces, significantly reducing analysis time and expediting decision-making. You can fuel blue-sky thinking sessions with business and finance stakeholders, enabling better decision-making for growth, new market penetration, new product pricing strategy, channeling the most profitable R&D investments and new product ideas. LLMs can help the treasury team optimize their forex cashflows by generating multiple hedging scenarios based on forex movement predictions using a combination of macroeconomic factors, external events, and exposure to multiple currencies.

4. Summarize Reports and Extract Insights: CFOs and finance teams are often swamped with lengthy macroeconomic reports, industry analyses, and market trends. Large language models can summarize this information for busy professionals, allowing CFOs to focus on the financial implications. The finance team becomes more analytical, dedicating their efforts to analyzing insights rather than generating them. E.g. the LLMs can help summarize compliance requirements per product line for each market, reducing the regulatory burden on the compliance teams.

5. Streamline Report Generation: AI empowers CFOs to automatically generate reports using a chat or voice interface. Imagine the convenience of generating reports in over 100 languages, even while on the move. This helps CFOs to make quicker, well-informed decisions.

6. ESG analysis & reporting: Finance teams are thinking about how to calculate their ESG metrics & optimize them. Most are struggling to find a starting point for their ESG initiatives. They can use AI models to identify key carbon consumption metrics and identify carbon-neutral strategies without impacting revenue & profitability.

Before embarking on your AI journey, do remember a few key considerations:

• Prioritize use cases based on business relevance and technological readiness. You can reference Gartner's prioritization model for initial guidance. I am building a more detailed prioritization model which I am happy to discuss with anyone interested.

Source: Gartner (October, 2022)

• ?Understand that LLMs work best when given the right context with an appropriate level of detail & guidance – (in other words, prompt engineering). Over time, finance teams should look to build a data bank of prompts & share them internally. Look at this video for examples of prompt engineering for finance users.?
• Keep data security and data sovereignty in mind when implementing AI models. Look out for an upcoming post dedicated to this topic.?
• Be aware of inherent biases in AI models, stemming from the training data and user biases. This topic also warrants a separate post.

In summary, AI models represent a groundbreaking evolution in finance. They have immense potential to revolutionize finance organizations and make them more forward-looking. AI could usher in a golden age for finance teams to rapidly become more strategic in nature by carefully selecting the right use cases and remaining vigilant about data security.

?

Disclaimer: The views expressed here are my own and do not necessarily reflect the views of Oracle.