The case for an AI monetary modeling agent: Cursor for Excel

Editorial Team
5 Min Read


The software program improvement world has been reworked by AI-powered coding assistants similar to Cursor and Claude Code, which have modified how engineers write and debug code. 

These instruments can generate contextually related code primarily based on person prompts, accelerating the software program improvement course of and enabling engineers to be a number of instances extra productive. Is it time for the same AI-powered monetary modeling assistant in Excel?

Monetary modeling, very similar to coding, includes quite a few interconnected spreadsheet tabs, complicated logical buildings constructed via Excel formulation and hyperlinks, and specialised area data. A great answer could be an AI modeling agent software the place it could actually perceive the context of spreadsheet tabs and generate fashions whereas monetary analyst prompts, very similar to a Cursor for Excel. Nonetheless, the present general-purpose instruments, like Excel Copilot, fall quick when requested to generate monetary fashions and we would want extra specialised options.

The demand throughout monetary disciplines

As a valuation analyst and former funding banking skilled, I incessantly use Excel to construct valuation fashions similar to Discounted Money Circulate and Choice Pricing Fashions. As an Excel energy person, I might eagerly subscribe to such an AI modeling software. The potential software of an AI monetary modeling agent spans throughout accounting companies, funding banks, and monetary and danger consulting companies, and so on. With a brief immediate, finance professionals can generate detailed fashions and collaborate with AI monetary modeling brokers all through the workflow.

The best way to construct the AI Monetary Modeling Software

I have not been in a position to uncover any comparable software presently out there available on the market. Probably the most simple strategy could be to develop an internet software as an Excel add-in, much like how Claude Code capabilities as an extension for VS Code. Alternatively, builders can select to construct an AI-native spreadsheet answer very similar to Cursor, the place each the normal spreadsheet and AI answer are built-in right into a unified platform. 

The distinction between coding recordsdata (e.g., .py) and Excel recordsdata (.xlsx) is perhaps that there is an additional visualization layer on prime of the code. This implies an AI modeling agent would possibly have to generate and execute code stay primarily based on human prompts, successfully simulating the modeling conduct of people in real-time. Happily, Excel supplied some scripting language instruments similar to Script Lab the place it could actually work together with Excel by way of APIs. Beneath is an experiment the place I attempted to create a mini discounted money circulation mannequin via prompting a Generative AI mannequin and Microsoft Script Lab.

DCF mannequin with Claude and Script Lab

The final setup concerned prompting Claude to create a mini discounted money circulation mannequin in Excel, then transferring the generated TypeScript and HTML code into the Script Lab Excel add-in for execution. The method seems to be like this:

First, I prompted Claude to create a DCF mannequin:

claude-dcf-model.png

Then, Claude generated TypeScript and HTML code containing the DCF mannequin logic and construction.

claude-typescript.png

In Script Lab, I pasted over and executed the AI generated code, which robotically generated a draft DCF calculation within the spreadsheet.

script-lab-claude.png

Admittedly, debugging the Script Lab code with Claude’s help took longer than constructing this straightforward mannequin from scratch. Nonetheless, as AI methods be taught and repeatedly self-train, I anticipate it might take approach much less effort to create a bigger and complicated mannequin throughout a number of spreadsheet tabs. Think about having a junior analyst who exists on a cloud server and may self-train monetary modeling 24/7? The productiveness features could be substantial.

Monetary modeling is overdue for its personal specialised AI assistant, very similar to how engineers now depend on instruments like Cursor. A devoted AI modeling agent may dramatically streamline the workflow for analysts, auditors and bankers alike. The chance is obvious: an AI-native monetary modeling assistant will rework how finance professionals work, and I am excited for the day when it turns into an indispensable a part of each finance skilled’s toolkit.

Share This Article