How to use Gemini Advanced for Finance?— The Ultimate Guide for Finance, Accounting and FP&A Teams

Congratulations!

You’ve got Gemini Advanced.

Now what?

How to use it for finance?

What are the best use cases for Accounting?

What can Gemini Advanced do that Copilot or ChatGPT can’t?

I’ll show you all of this

What is Gemini Advanced?

Gemini Advanced is a cutting-edge, large language model (LLM) developed by Google AI. It stands out for these reasons:

• Powered by Ultra 1.0: Unlike many LLMs, Gemini Advanced leverages Google’s Ultra 1.0 model, resulting in superior reasoning, complex task mastery, and deeper understanding and generation of language.
• Contextual Understanding: Gemini Advanced excels in grasping the nuances of financial and accounting data. It processes details with exceptional efficiency.
• Learning and Adaptation: As you work with it, Gemini Advanced learns your preferences and workflows, continuously refining its outputs and suggestions.

How to Use Gemini Advanced for Finance

Here are several ways finance professionals can use Gemini Advanced:

Financial Report Writing: Streamline the process of creating reports by asking Gemini Advanced to provide insightful summaries of financial data.

Additionally, use the model to quickly format and structure reports according to specific standards.

2. Variance Analysis: Uncover the causes of budget or forecast variances using Gemini Advanced.

Input your data, ask targeted questions and discover underlying factors influencing business results.

3. Budget and Forecast Refinement: Gemini Advanced can assist in forecasting scenarios by analyzing trends, incorporating economic insights, and offering suggestions on potential adjustments. This makes your budgets and forecasts more robust.

4. Automated Account Reconciliations: Reconciling accounts is often tedious.

Gemini Advanced analyzes large datasets with speed and accuracy, comparing entries for potential discrepancies and providing reconciliation assistance .

It gives you great Python code for this:

Below is the code in case you want to test it yourself.

import pandas as pd

# Load data (Assume Excel for illustration)
bank_data = pd.read_excel("bank_statement.xlsx")
accounting_data = pd.read_excel("accounting_records.xlsx")

# Basic Matching 
matched_txns = pd.merge(bank_data, accounting_data, on=['Date', 'Description', 'Amount'], how='inner')

# Unmatched in Bank Statement
unmatched_bank = bank_data[~bank_data.isin(accounting_data).all(axis=1)]

# Unmatched in Accounting Records
unmatched_accounting = accounting_data[~accounting_data.isin(bank_data).all(axis=1)]

print("Matched Transactions:\n", matched_txns)
print("Unmatched in Bank:\n", unmatched_bank)
print("Unmatched in Accounting:\n", unmatched_accounting)

If you have read my previous articles, you know I am big advocate to use Google Colab for Finance.

With Gemini advanced, you can even export the code directly to Colab.

And it creates a notebook for you:

Top Accounting Use Cases

1.Tax Code Analysis: Gemini Advanced delves into tax codes and regulations, generating comprehensive summaries.

With its understanding of context, it clarifies complex language and regulations with remarkable ease.

Prompt Used: Analyze the Netherlands tax code in relation to VAT on cross-border transactions with Belgium. Focus on applicable rates, rules for determining the place of supply of services, and any relevant exemptions. Summarize the key points I need to be aware of in plain language.

How Gemini Advanced Makes This Easier

• Distills Complexity: The Netherlands tax code is massive and filled with legal detail. Gemini Advanced cuts through this, extracting the most relevant information.
• Plain Language: It avoids dense jargon, making the tax implications easier for a non-expert to understand.
• Highlights Action Points: Gemini Advanced doesn’t just give information, it hints at next steps (like seeking specialist advice).

2. Audit Assistance: During audits, the model can assist in extracting critical information from large financial datasets, providing targeted summaries from extensive documents for increased efficiency.

This is how Gemini Advanced could provide audit assistance.

Here’s a scenario demonstrating its capabilities:

Scenario: You’re an auditor reviewing a company’s travel and expense (T&E) policies and supporting documentation. You need to verify compliance with internal rules and identify potential areas of high risk for abuse.

Prompt for Gemini Advanced: “Analyze this dataset of expense reports [provide data, either a file or a description of its structure].

Focus on the following:

• Summarize the top 5 expense categories with the highest total amounts.
• Flag any expense submissions exceeding €500 for further review. Identify employees with more than 3 expense reports submitted within a single week.
• Are there expense descriptions missing sufficient detail or using vague terms (e.g., “miscellaneous”)? List those.
• Cross-reference a random sample of 10% of expense receipts against a list of approved vendors. Highlight any discrepancies.”

And the response would be something like this:

3. IFRS vs GAAP: If you face questions about Accounting standards, Gemini can help you out!

This is how Gemini Advanced could assist with understanding accounting standards like IFRS and GAAP.

Scenario: You’re working on a financial reporting project for a multinational company. You need to ensure compliance with IFRS, but are more familiar with US GAAP and unsure about key differences in inventory valuation.

Prompt for Gemini Advanced

“Explain the major differences between IFRS and US GAAP regarding inventory valuation. Provide practical examples to illustrate the impact on financial statements.”

And this would be the answer:

Gemini Advanced vs. ChatGPT and Copilot

It is too early to tell but I will mention some differences that I have found.

Google Ecosystem Integration: An advantage (especially with extensions) is Gemini Advanced’s integration with Google tools (Google Colab, Gmail, Sheets, etc.).

This offers potential for smoother workflows where you can access and generate financial insights directly within the tools you already use.

Comparison betweek Models: Now, this information below is directly from the Gemini website so it might be biased in favour of Gemini.

These table in the image shows how well each model performed on a variety of tasks, such as answering questions about different subjects, doing math problems, and generating Python code. In almost all of the tasks, Gemini Ultra outperformed GPT-4, although the difference was sometimes small.

Overall, the image suggests that Gemini Ultra is a more powerful and versatile LLM than GPT-4. However, it is important to keep in mind that these models are still under development, and their performance can vary depending on the specific task.

50 Ways to use Gemini in Finance, Accounting and FP&A

Ready to put Gemini Advanced to work? Experiment! Begin with straightforward tasks and gradually increase complexity.

Ask for summaries, reports, or analyses specific to your role and industry.

The more you integrate, the more valuable Gemini Advanced becomes.

Here are 50 prompts tailored for finance, accounting, and FP&A professionals.

I am going to try them in my next few articles.

Financial Analysis & Reporting

1. “Summarize these key trends from our income statement over the past three quarters.”
2. “Compare the profitability ratios of our company to the industry average. Provide insights.”
3. “Draft a memo explaining the reasons behind the significant increase in operating expenses.”
4. “Create a chart visualizing our cash flow projections for the next 12 months.”
5. “What are the top three factors likely to impact our revenue growth in the upcoming quarter?”
6. “Analyze this customer data to identify high-churn risk segments.”
7. “Provide a risk assessment of our current working capital position.”

Budgeting & Forecasting

1. “Generate a sales forecast based on historical data, seasonality, and these potential market trends.”
2. “Develop three budget scenarios: optimistic, baseline, and pessimistic. Highlight key assumptions.”
3. “Calculate the potential ROI of investing in this new equipment. Include time value of money.”
4. “How would a 5% increase in raw material costs impact our profit margins? Model this for me.”
5. “Adjust this budget model to reflect a potential headcount reduction in Q3.”

Accounting

1. “Explain the double-entry accounting principle for this complex transaction, with examples.”
2. “Are these transactions compliant with GAAP/IFRS? Cite relevant sections of the standards.”
3. “Classify these expenses according to their tax deductibility.”
4. “Write a journal entry to record the depreciation of this asset using the straight-line method.”
5. “Help me reconcile this bank statement. Here’s the data — look for potential discrepancies.”

Python-Specific Prompts

1. “Generate Python code to calculate the weighted average cost of capital (WACC).”
2. “Write Python code to scrape financial data from [website] and import it into a DataFrame.”
3. “Build a Python function to automatically calculate key financial ratios from a given dataset.”
4. “Using this sales data, write Python code to create a customer segmentation model.”
5. “Generate Python code to perform Monte Carlo simulation for this investment risk assessment.”
6. “Create a Python script to automate the generation of our monthly financial reports.”
7. “Help me optimize this Python code for calculating inventory turnover. Are there efficiency gains?”
8. “Build a Python function to perform a DuPont analysis on this financial data.”

Tips

• Clarity is key: The more specific and detailed your prompts, the better the results.
• Start simple: Begin with basic tasks and gradually increase complexity.
• Experiment: Test different variations of your prompts to discover what works best.