Hi Reader,

These techniques aren’t new for me.

I’ve been teaching them in my classes for a while now, and they’ve become the core of how hundreds of finance pros are using AI to work faster, think deeper, and deliver insights that stand out.

But I realized many of you reading this newsletter haven’t seen them all in one place.

If you’re getting started with AI or is already using ChatGPT, Copilot, or Gemini...but your prompts feel hit-or-miss, this will fix that.

But before I break it down, one quick thing...

24 Hours Left to Enroll: AI for Finance - Advanced

This is your chance to join a small, elite cohort of finance professionals learning how to use ChatGPT, Copilot, Gemini, and Python in finance on an advanced level.

​👉Save $100 and secure one of the last spots now👈 ​

Here's what you'll learn:

• Give advanced instructions to Generative AI
• Use AI for Financial Analysis (with real data)
• Solve the Confidential Data Problem
• Build your own GPT for Finance
• Use Python + AI (even if you’re a beginner)

New this live course:

• Full training on GPT Builder — how to create your own GPT with internal data
• A step-by-step Python Learning Track + 31 finance-focused notebook exercises
• Make AI build a super visual report in 5 minutes

👉Enroll now before registration closes tomorrow​
(The coupon code with the $100 discount is applied in the link)

11 Finance Prompting Techniques You Should Try Today

Below are the exact prompt engineering methods I use daily and that I teach live in my trainings and workshops.

I have made sure to go directly to practical information and prompt examples.

Enjoy!

1. Basic Prompting – CSI + FBI Framework

Most people write vague prompts. That’s why they get vague answers.

Use this formula instead:

• CSI: Context + Specific + Instruction
• FBI: Format + Blueprint + Identity

Example:​
“I'm a consultant. My client has an overdue invoice for 2 months. Can you draft a formal communication to recover the payment? The communication needs to be a formal letter. I want the communication to be hard and use words like legal actions. Draft this letter like you would be the best lawyer.”

This prompt works because it tells the AI:

• Who you are
• What you need
• How it should sound
• And in what format

When to use it:​
Literally on any general task, from overdue payment emails to CEO briefings to summary reports, etc.

2. Chain-of-Thought

Turn big problems into bite-sized steps. Works wonders for reconciliations, forecasts, or policy modeling.

How to do it:

1. Identify the core question
2. Break it down
3. Ask sequential questions
4. Guide the reasoning
5. Synthesize the conclusion
6. Review and refine

Example use case:​
"Help me automate the reconciliation of vendor invoices from multiple systems using a step-by-step logic.”

When to use it:
​When tackling intricate tasks like automated reconciliations, policy building, or multi-step variance analysis.

3. Chunking

AI can get overwhelmed by large blocks of text or data. So give it smaller chunks. Ideally after the first prompt, you add a follow-up.

Use chunking for both input & output:

• Break complex variance reports into sections
• Ask ChatGPT to output one segment at a time

Example:​
“Provide a structure for an annual financial report for an SME, breaking it down into key sections with a brief description of what should be included in each.”

When to use it:​
When working with long policies, big data summaries, budget variance explanations, or board prep decks.

4. Explicit Reasoning

Ask the AI to explain its logic, just like a finance analyst would.

Prompt example:​
“Using the following data, calculate the three most important liquidity KPIs for our company. Please provide a step-by-step explanation of each calculation. Assume our current assets are $500,000, inventory is $150,000, current liabilities are $250,000, and cash & cash equivalents are $200,000.”

Perfect for liquidity modeling, ROIC breakdowns, or debt service coverage ratios.

When to use it:​
Ideal for KPI calculations, liquidity models, cash flow diagnostics, and metric deep-dives.

5. Meta-Cognition

Want better answers? Ask the AI to reflect on its own reasoning.

Prompt example:​
"Evaluate the reliability of your forecast based on assumptions. What could go wrong?”

This helps stress-test scenarios and explore the limitations of AI-generated insights.

When to use it:​
Useful when refining forecasts, stress-testing strategies, or improving data quality reviews.

6. Socratic Prompting

Ask open-ended questions. Challenge assumptions.

This method unlocks more depth in strategic planning and scenario design.

Prompt Example:​
“As a CFO advisor, help me think through the implications of delaying vendor payments to improve short-term cash flow.

• What are the potential long-term consequences of this strategy beyond immediate liquidity relief?
• What assumptions are you making about supplier relationships and their flexibility?
• How would this strategy impact our working capital and overall financial health over the next two quarters?
• Can you explain any ethical or reputational risks that might be involved here?

What alternative strategies could improve cash flow without risking operational continuity or partnerships?”

When to use it:​
For strategy development, scenario design, or internal audit prep.

7. Agent Prompting

Don’t just ask. Assign a role.​
Tell the AI: “You’re an FP&A Analyst…” and then give context.

Define these 9 traits:

1. Name
2. Definition
3. Knowledge
4. Traits
5. Analysis
6. Output
7. Format
8. English level
9. Start

Example prompt:​
"You are a seasoned Corporate FP&A Analyst. Your task: Create a liquidity dashboard using data from the cash flow forecast and bank statements.”

When to use it:​
Use it for management reporting, scenario analysis, or financial modeling.

8. Team Prompting

Simulate a team conversation by combining multiple “agents.”

Use Case:​
Get input from Finance + Ops + Marketing in one thread for more realistic scenario planning or product P&L design.

Prompt example:​
“Finance: Estimate margin impact.
Ops: Identify resource constraints.
Marketing: Forecast campaign uplift.”

When to use it:​
For cross-functional planning, P&L assessments, or strategic brainstorming.

9. Iterative Inquiry & Sequential Questioning

Don’t ask for everything in one go. Ask one layer at a time.

Prompt:​
“I need help with this forecast. Can you check if the assumptions are reasonable before we go into the model?”

Then:
“Now, what would you change in the cost structure to reflect a 10% price drop?”

This method helps for deep dives, especially in Excel-heavy workflows.

When to use it:​
Crucial for decision support, forecast refinement, or progressive spreadsheet building.

10. Fact Checking

Stay credible.

Prompt:​
“Provide 3 sources to support your answer. Include publication name, date, and URL.”

Use this when drafting investor memos or board packs to make sure what you say stands up to scrutiny.

When to use it:​
Great for board materials, investor decks, regulatory summaries, or anything that needs to be 100% reliable.

11. Prompt Optimization & Expansion

Already have a prompt?

Let the AI improve it.

Prompt:​
“This is my draft prompt. Make it more precise and effective for a financial modeling use case.”

You’ll get more structured, reliable, and insightful answers, with zero time wasted starting from scratch.

When to use it:​
Whenever you want to refine your results without starting over, like dashboards, emails, or data requests.

Have you already tried any of these techniques, but you weren't aware you were actually using them?
​
Let me hear your feedback and learn how I can help you further in my next editions.

​

Best,
Your AI Finance Coach, Nicolas
​
​
​P.S. This course is the only place where we teach our proprietary framework for using AI with your actual data without worrying about data confidentiality.
​ 👉 Reserve one of the last spots now and secure your discount​