Automation Bot - Payroll and Record keeping

The Objective

A contact who owns a limousine business was spending days manually scraping data and building spreadsheets to calculate employee payroll each month. I built a Python web scraper to parse the DOM of his 3rd-party CRM, extract the relevant metrics, and pipe them directly into a Google Sheet via API. I then built a data pipeline using Pandas to process the raw numbers into a structured report, which is fed into an LLM. Now, he can simply ask the bot, 'How much did driver X make this month?' and get an instant, accurate answer. Processing the data with Pandas first was a critical design choice—dumping raw tabular data directly into the LLM caused severe hallucinations, so the Python backend handles the deterministic math while the LLM handles the natural language interface.