Leveraging Generative AI through Effective Prompt Engineering

“We’re no longer in the information age, but [really] a conceptual age. And I know people like to talk about AI as a fourth industrial revolution. But I [actually] think it’s more of a concept of curation.” —Danielle Supkis Cheek, VP and Head of Analytics and AI

As companies increasingly adopt AI-driven solutions to streamline operations, it’s clear that GenAI offers a path to greater efficiency, accuracy, and insight. According to a recent KPMG survey, 99% of companies globally plan to pilot or actively implement AI for financial reporting within the next three years, with 75% of firms in Australia already integrating AI into their financial processes. With these statistics in mind, it’s evident that GenAI has the potential to reshape audit and accounting functions. Still, the effectiveness of this transformation hinges significantly on how AI is utilised—particularly through prompt engineering.

The growing role of AI in audit and accounting

AI adoption in financial reporting and auditing is on a sharp rise. As KPMG reports, companies are setting aside significant portions of their IT budgets to fund AI initiatives—10% on average worldwide, with half of these companies expecting a further increase in investment by 2025. This commitment reflects AI’s anticipated impact, as companies recognise that GenAI not only improves efficiency but also strengthens analytical capabilities. In Australia, 51% of surveyed companies reported spending between 11% and 20% of their IT budgets on AI.

This rapid increase in AI adoption is driven by its potential to automate labour-intensive processes, enabling auditors to focus on high-value tasks that improve audit quality. Deloitte’s research supports this, showing that GenAI automates repetitive tasks, improving audit quality and efficiency. Furthermore, boards are increasingly urging auditors to apply AI to identify anomalies and assess risk, with 73% of board members expecting auditors to prioritise these capabilities.

The importance of Prompt Engineering in maximising AI utility

To truly harness GenAI’s capabilities in the audit and accounting sectors, companies need to adopt effective prompt engineering practices. Prompt engineering—the practice of designing and refining input prompts to achieve desired AI outputs—is fundamental to maximising AI’s performance and aligning it with specific business objectives. Here’s how prompt engineering can optimise AI utility in audit and accounting:

• Refining anomaly detection

To ensure AI accurately identifies anomalies in financial data, auditors can use prompts that set explicit parameters for risk tolerance, data thresholds, and industry-specific metrics. A well-crafted prompt might ask the AI to “analyse year-over-year revenue trends and flag any deviations exceeding 5% for further review.” This tailored prompt directs the AI to focus on relevant patterns, ensuring more precise anomaly detection.

• Predictive analysis

As KPMG highlighted, 73% of board members prioritise AI’s ability to perform predictive analysis. Prompt engineering allows auditors to guide AI towards forecasting future trends and potential risks. For instance, a prompt might be, “Based on the past five years of financial data, predict revenue growth for the next two quarters and identify potential risks impacting these projections.” This approach helps focus the AI’s analysis on specific timelines and risks, yielding actionable insights.

• Enhancing financial reporting accuracy

GenAI can enhance accuracy in financial reporting by automating data analysis and reconciliation tasks. However, to leverage this effectively, prompt engineering should involve clear and direct instructions. For instance, a prompt might be, “Reconcile discrepancies between income statements and balance sheets, identifying potential data entry errors.” This prompt directs the AI to perform targeted tasks, ensuring accuracy and consistency across financial documents.

• Creating auditable reports

AI’s ability to generate reports that are easy to audit can be amplified through prompt engineering. For example, prompts that include language such as “Generate a report summarising quarterly revenue with a breakdown by region and highlight any inconsistencies between this and the last quarter” can streamline report generation. By guiding the AI to focus on specific components, prompt engineering helps produce outputs that are both informative and structured for easy auditing.

Balancing risk management with AI utility

While GenAI presents immense opportunities, it also introduces risks that must be managed carefully. According to the Centre for Audit Quality (CAQ), the use of GenAI in auditing brings potential risks related to data accuracy, confidentiality, and security. This is where prompt engineering plays a dual role—not only guiding the AI to generate useful information but also ensuring that outputs align with compliance and ethical standards. For instance, prompts can specify that AI limit its scope to anonymised datasets, or focus only on approved metrics and parameters.

The future of audit and accounting with GenAI

As more companies integrate GenAI into their financial operations, the role of prompt engineering will continue to grow. Effective prompt engineering enables organisations to refine AI outputs and maintain quality, accuracy, and compliance standards. For companies aiming to maximise their AI investment, a structured approach to prompt engineering can streamline the process, reducing repetitive work and producing more reliable insights.

To support your organisation’s journey in adopting AI for audit and accounting, don’t miss Danielle Supkis Cheek’s insightful session on “Embracing GenAI and the Power of the Prompt“ from CwX APAC 2024. Explore practical strategies to craft effective prompts, learn best practices, and unlock the transformative potential of AI in your workflows. Watch here.