How we integrate AI tools into our financial modeling process—and how you can too
Financial modeling is a necessary task that companies eventually have to tackle, whether they’re just starting up or already generating millions in revenue. Sooner or later, they’ll need comprehensive financial analysis. In recent years, a common question has emerged: Can we effectively use AI for financial analysis?
If you Google this question, you’ll find numerous services claiming to offer AI for financial analysis. But there’s a significant difference between services that simply claim AI capabilities and those that actually deliver useful results.
As a Swiss business consultancy focused on investment preparation, we’ve developed numerous financial models for our clients. Here’s our practical guide to how we integrate AI for financial analysis into our workflow—and how you can do the same.
Our AI-Assisted Financial Modeling Process
Our models can be as sophisticated as three-statement models for companies with multiple subsidiaries. We’ve successfully incorporated AI for financial analysis using a straightforward approach:
1. Flow Charting: The Critical First Step
Before we even open Excel, we create a detailed flow chart of the entire model—essentially designing the spreadsheet architecture from scratch. This visual map shows which elements connect to others and how information flows through the financial model.
For this flow charting phase, we typically use tools like:
- Miro
- Apple’s Freeform
- Any software that offers robust flow charting capabilities
This visual representation helps us refresh our thinking about how different components connect. The flow chart places all the ground rules in one place, ensuring we don’t overlook critical elements related to revenue, expenses, or other financial factors.
How AI helps with flow charting: This is where AI for financial analysis first comes into play. We use Anthropic’s Claude, which has excellent capabilities for helping develop flow charts. Claude can suggest elements we might have overlooked and help visualize the connections between different components of the financial model.
Pro tip: Don’t rush this planning phase. Thoroughly review your flow chart and double-check that you haven’t forgotten any elements related to revenue or expenses—such as legal costs or technology investments.
2. Setting Up an AI Project Environment
Once we have our flow chart, we establish a dedicated project in Claude (although GPT-4 can also work well). We might name it something like “Financial Model for [Company Name]” and upload all the relevant information the AI might need.
We then add our flow chart to this project, ensuring the AI understands the overall structure we’re working toward. This creates a “knowledge base” for the AI about our specific project before we even begin building in Excel.
Why we prefer Claude: While both Claude and GPT-4 can handle these tasks, we’ve found Claude’s project integration particularly useful. The ability to maintain context throughout a complex financial modeling process gives it an edge for our workflow.
3. Building the Model with AI Assistance
Here’s where expectations need to be managed. Current AI for financial analysis isn’t at the point where you can simply say “create me a complete financial model” and get perfect results. Instead, we use AI as an execution assistant and thought partner.
Our process looks like this:
- Ask the AI which sheets/tabs we should create in Excel
- Request specific formulas and equations for different sections
- Implement the AI’s suggestions manually in Excel
- Take screenshots of our work when we need clarification
- Ask the AI to review our implementation and suggest improvements
The screenshot approach helps overcome Claude’s inability to directly interact with Excel. While the AI might sometimes reference incorrect cell addresses, understanding the logic behind the formulas allows us to adjust them appropriately.
Current Limitations of AI for Financial Analysis
It’s important to acknowledge that this approach isn’t perfect:
- AI tools can’t directly manipulate Excel (yet)
- They sometimes reference incorrect cell addresses
- Complex financial modeling still requires human oversight
- You need to understand finance fundamentals to evaluate AI suggestions
While there are online applications like QuickBooks that offer some integration with financial analysis tools, none have proven sophisticated enough for complex financial modeling as of now.
Why This Approach Works for Us
Despite these limitations, we’ve found significant value in using AI for financial analysis:
- Time savings: AI can instantly suggest solutions to specific financial modeling challenges that might otherwise take hours to resolve
- Reduced isolation: Financial analysis can be solitary work—having an AI assistant provides a thought partner
- Error reduction: AI can spot inconsistencies in formulas or assumptions that humans might miss
- Knowledge enhancement: Working with AI often teaches us new approaches to financial modeling
Getting Started with AI for Your Financial Analysis
If you want to implement a similar approach, here’s how to begin:
- Start with a clear understanding of your business model and financial structure
- Create a detailed flow chart of your intended financial model
- Set up a project in Claude or similar AI tool
- Use the AI as a guide rather than expecting it to build the entire model
- Implement the AI’s suggestions manually while applying your own financial expertise
- Use screenshots to get AI feedback on your implementation
Remember that the goal isn’t to replace your financial expertise but to enhance it. AI for financial analysis works best as an assistant that helps you create not just a good model, but one that you thoroughly understand.
Conclusion
As financial modeling evolves alongside AI capabilities, we expect these tools to become increasingly powerful. For now, our approach of using AI as a guided assistant rather than a replacement for financial expertise has proven highly effective.
Give it a try—set up Claude as your assistant for financial analysis, and you’ll likely find yourself creating more robust models in less time, all while deepening your understanding of the financial structures you’re modeling.