Harnessing Generative AI for Enhanced Accounting and Auditing

For the past 10+ years, finance-oriented data scientists have been perfecting machine learning using artificial intelligence. Many of these financial data scientists were trained at companies like American Express and Capital One, where they used huge datasets of consumer data to determine the creditworthiness of prospective borrowers.

The Evolution of Financial AI

Traditional Regression Analysis

It started with traditional logistic regression analysis, where data scientists would regress one, two, or three variables against a dataset to determine which variables were most statistically significant when predicting the likelihood of default. They would analyze variables like:

• Home ownership
• Car ownership
• Number of credit cards outstanding
• Total debt-to-income ratio
• FICO score

The Machine Learning Revolution

When machine learning emerged, the game changed dramatically. Data scientists upgraded from regressing a couple of variables to regressing all variables against all other variables in every combination possible—oftentimes resulting in tens or hundreds of millions of combinations.

They used deep learning techniques to train neural networks with multiple layers of data to understand relationships between all variables. This type of analysis was beyond human capacity and required machines to complete the computation. It was the dawn of machine learning.

Personal Experience in FinTech

In 2011, I co-founded a credit decisioning business called LendingRobot (originally Lend Academy Investments), which applied logistic regression and neural networking to datasets from P2P lenders like LendingClub and Prosper. We:

• Hired data scientists from big firms
• Innovated new ways to analyze data
• Created credit models more predictive than legacy models
• Helped expand access to capital for qualified borrowers

Machine learning yielded amazing results in credit decisioning, allowing lenders to expand their set of qualified borrowers while minimizing default rates. This was early financial AI—my first experience using this technology.

The Generative AI Revolution

Machines Master Human Language

We are now entering a new phase of AI with far larger implications for society. In this new era of Generative AI, machines have mastered human language. This is about organizing unstructured data, analyzing, and predicting based on qualitative data.

While the technology isn't perfect, it's very impressive at this early stage and will continue to get smarter rapidly. Already, it has an amazing ability to read, write, speak, and reason, adjusting for tone, inflection, and personality.

The Transformer Breakthrough

After the invention of transformer technology in 2017—the underlying neural network that powers large language models like OpenAI's GPT-4 Turbo—processing huge volumes of qualitative data like text, images and videos became possible.

Transformer technology utilizes an AI concept called "attention" to emphasize the weight of related words, providing context for words or tokens. Data scientists have utilized this technology to "compute language," enabling machines to:

• Convert meaning and context of words into weighted, associated data
• Produce statistical probability of the next word in a sentence
• Generate new word clusters

When ChatGPT launched in late 2022, it demonstrated to the world that Generative AI is natural, intuitive, and intelligent.

Generative AI as a Paradigm Shift

"Generative AI is as revolutionary as mobile phones and the Internet"
— Bill Gates

Over the past 40 years, we've experienced technological breakthroughs that propelled technology forward:

• The microprocessor
• Graphical user interface (GUI)
• Browser-based Internet
• Mobile technology (putting computers in our pockets)

Generative AI is the latest breakthrough technology. While we've experienced Siri, Alexa, and Google Assistant, ChatGPT demonstrated the major next step—comprehensive responses that are easy to understand.

Over the next few years, there will be a huge wave of new natively built Generative AI applications that will unlock power in ways we haven't dreamed of yet.

Tellen's Focus: Reinventing Accounting

My co-founders and I have committed ourselves to reinventing accounting using Generative AI. We want to build a native Generative AI application that wasn't possible before this paradigm was introduced.

Why Financial Audits?

Given our deep experience building fintech applications, we chose financial services as our vertical. We looked for work functions that require knowledge workers who organize, analyze, and predict based on large volumes of quantitative and qualitative data.

We landed on accounting and specifically financial audits as an ideal use case for Generative AI.

The US Audit Industry by the Numbers

• $144B total US accounting industry revenue in 2022
• $60B from audit services specifically
• 46,000 accounting firms in the US
• $12M average S&P 500 company audit cost
• 30,000 hours required for average S&P 500 audit
• 17% decline in number of accountants over past 2 years
• All-time high number of accounting job openings
• 40% of all audits have deficiencies according to PCAOB

Market Research Insights

We've conducted more than 90 product research calls with accounting professionals. Our takeaway: Generative AI is a high priority for accounting firms, and their first step is often launching a private instance of an LLM that is secure and confidential for firm and client data.

Tellen's Solution Architecture

The Audit Data Challenge

When an auditor renders judgment on financial statement validity, they utilize three sets of data:

1. Publicly available data (GAAP, IFRS, Tax Code, etc.)
2. Firmwide methodologies (typical Big 4 audit methodology handbook: ~7,000 pages)
3. Client data (must be validated)

Audits take tens of thousands of hours and cost millions of dollars because they require meticulous analysis and reconciliation of huge amounts of unstructured data (bank statements to invoices, proper disclosures, risk standards, referenced numbers with cited evidence).

Our Technical Approach

Tellen's audit management solution generates contextualized accounting-related answers by optimizing LLMs through:

• Prompt engineering
• Retrieval-augmented generation (RAG)
• Fine-tuning

This generates answers specifically tailored to our industry, producing higher quality responses than asking the same questions directly to an LLM.

Product Roadmap

Phase 1: Tellen's accounting chatbot (currently available) Phase 2: Enterprise audit management system with proprietary data integration Future: Full-suite AI-powered audit management system

Expected Impact

We believe our solution will:

• Significantly increase auditor productivity
• Increase audit quality
• Pass along margin improvements to accounting firms and their clients
• Enable auditors to take on more clients and render judgment based on better data
• Help accounting firms accelerate revenue AND expand EBITDA margins by 25% or more

In an era where accountants are in short supply and audit quality is increasingly deficient, our solution cannot arrive at a better time.

Get Started Today

We encourage anyone in the accounting industry to use our accounting chatbot and share it if helpful. We'll refine the chatbot based on your feedback and use learnings from this free product to inform development of our enterprise audit management system.

Try Tellen's accounting chatbot today and help us build the future of AI-powered auditing.