Overview

New Clarity designed and implemented a custom Model Context Protocol (MCP) server in Python to streamline one of the financial workflows for a fractional CFO. The solution connected directly to Claude desktop as the LLM client, enabling the AI to intelligently transfer financial data between spreadsheets. This project reduced hours of manual work by automating the matching and population of data fields across models.

The Challenge

Finance teams regularly work with profit and loss (P&L) statements and separate financial models. Although both contain related data, the fields are often named differently, forcing analysts to spend time manually aligning numbers. This repetitive task created a risk of human error and consumed valuable time that could be spent on analysis rather than data entry.

The New Clarity Solution

To address the problem, New Clarity built an MCP server in Python that connected to Claude desktop. This allowed the LLM to act as an intelligent intermediary whenever a user needed to transfer data from one spreadsheet to another.

The process worked as follows:

• Data Extraction: The MCP server read columns and values from both the P&L spreadsheet and the target financial model on the local file system.
• Context Delivery: Both sets of column data were provided to Claude for analysis.
• Intelligent Mapping: The LLM compared the fields and calculated probability scores for matches, even when naming conventions differed.
• Automated Transfer: Based on the highest-probability matches, the MCP server moved values into the correct fields of the financial model.
• User Oversight: Finance professionals could review or adjust mappings if needed, but the system handled the majority of the work automatically.

Results

By combining MCP with Claude desktop, finance professionals gained a powerful new workflow:

• Time Savings: Analysts no longer had to manually map and transfer each line of data, saving hours of repetitive effort.
• Error Reduction: AI-based matching reduced the risk of mistakes caused by human oversight.
• Workflow Efficiency: The process of building financial models became faster and smoother, allowing teams to focus on higher-value strategic work.

‍

“The enhancement built directly into Claude desktop is used to manipulate our spreadsheets to build financial models from P&L statements and other documents. This has saved me countless hours of time I would have otherwise spent doing manual entry, so now I can now allocate that time to more valuable work.”

– Rob K, Fractional CFO

Why It Works

Traditional automation tools struggle when data fields do not align perfectly. By leveraging an LLM, New Clarity was able to apply contextual reasoning to match fields based on meaning rather than exact names. The MCP server created a seamless bridge between the local file system and the LLM client, allowing complex data transfers to be executed reliably and repeatedly.

Next Steps

This project illustrates how AI agents can extend far beyond simple chat interactions. Similar MCP-based workflows can be developed for any domain where unstructured or inconsistently labeled data must be intelligently aligned, such as supply chain records, HR systems, or legal documents. New Clarity continues to help organizations design and deploy agents that remove repetitive work and unlock greater efficiency.

‍