JAN–MAR 2026 · FINANCIAL MODELING
FpInvest
Financial modeling from company earnings reports
Python · LangChain · knowledge graphs · Excel
FpInvest, formerly FirstPrinciples.AI, was a platform for updating financial models using company earnings reports. These models use financial data and assumptions to estimate a company’s performance.
Built the Python/LangChain workflow connecting report parsing, company dependencies, and updates to Excel models.
Connecting earnings reports to company models
Built a Python and LangChain workflow to parse earnings reports. Used knowledge graphs to represent dependencies between companies and propagate changes through those relationships.
The dependency model mattered because a changed assumption could affect more than one company’s model. Representing those relationships explicitly gave the workflow a way to carry a change through connected companies, beyond extracting figures from a single report.
Updating the spreadsheet
The output needed to reach the financial model an analyst worked with. Exported the updates to Excel and kept a record of what changed, connecting the report-processing workflow to spreadsheet-based analysis.