Gravitas Fpa Agent

QSR FP&A Analysis Agent

A simple but real FP&A analysis agent built in Python and Jupyter. It ingests QSR store sales data, surfaces revenue variances, margin anomalies, and labor cost issues, then calls the Claude API to generate a plain-English executive summary — the kind of output a CFO or PE Operating Partner would actually read.

Built as a portfolio project to demonstrate applied AI thinking in an FP&A and QSR context.

What It Does

1. Loads a CSV of QSR store-level sales data (actual vs. forecast)
2. Calculates revenue variance, gross margin, and labor cost ratios per store
3. Flags underperforming stores and anomalies against QSR benchmarks
4. Calls the Claude API with the analysis context
5. Outputs a concise executive insight summary with prioritized recommendations

Sample Output

**[View the full notebook with live output →](https://nbviewer.org/github/adixit22/gravitas-fpa-agent/blob/main/notebooks/fpa_agent.ipynb)**

**[Read the full executive report →](sample_output/executive_report.txt)**

Store Performance Summary (Jan–Feb 2025)

Store	Region	Avg Rev Variance	Avg Gross Margin	Avg Labor Ratio	Total Actual Rev
Airport Terminal	East	-28.9% ⚠️	56.0%	30.0% ⚠️	$299,000
Industrial Zone	East	-25.5% ⚠️	52.0% ⚠️	30.0% ⚠️	$210,000
Highway Junction	North	-25.3% ⚠️	55.0%	30.0% ⚠️	$263,000
Riverside Park	North	-9.9%	55.0%	30.0% ⚠️	$299,000
Downtown Core	East	-5.1%	60.0%	25.0%	$370,000
University District	West	-0.9%	58.8%	25.0%	$359,000
Suburban Plaza	North	+2.0%	60.0%	25.0%	$410,000
Lakeside Drive	East	+2.6%	61.0%	25.0%	$443,000
Westside Mall	West	+6.7%	60.0%	25.0%	$496,000
City Centre	West	+7.9% ✅	62.0% ✅	25.0%	$545,000

*QSR benchmarks: Gross Margin ~60% | Labor Ratio 25–28% | Revenue Variance ±5% acceptable*

Executive Insight (Claude-generated)