GenAI Reporting Pipeline

Replaced a 5-stage manual sales reporting process with a single command.

	Before	After
Step 1	Download CSV, open Excel	python genai_pipeline.py
Step 2	Clean & filter data manually	Pandas validation + enrichment
Step 3	Calculate KPIs in formula cells	Python metric engine
Step 4	Check for anomalies by eye	Rule-based anomaly detection
Step 5	Write summary email to leadership	Gemini 1.5 Flash executive summary

Result: One command. Zero manual steps. Full stakeholder report in seconds.

---

How it works

data/sales_data.csv
       ↓
  Validation & enrichment (Pandas)
       ↓
  KPI computation (revenue, returns, MoM growth, top performers)
       ↓
  Anomaly detection (statistical flagging)
       ↓
  Executive summary (Gemini 1.5 Flash — free tier)
       ↓
outputs/report_YYYYMMDD.md  ← ready to share

See a real output: outputs/sample_report.md

---

Automation (Zapier equivalent)

Drop a CSV into a watched folder → pipeline fires automatically.

New CSV in folder
      ↓
pipeline_runner.py detects it (watchdog)
      ↓
genai_pipeline.py runs automatically
      ↓
Report saved + email notification sent

Full Zapier setup guide → automation/zapier-docs/ZAP_SETUP.md
Coded Python fallback → automation/pipeline_runner.py
Make (Integromat) blueprint → automation/zapier-docs/make_scenario_export.json

Live trigger log (from actual test run):

triggered_at	file	status	duration
2026-06-13 06:53:08	sales_jan_2024.csv	SUCCESS	0.5s

---

Setup

git clone https://github.com/tanishaa11/genai-reporting-pipeline
cd genai-reporting-pipeline
pip install -r requirements.txt

# Get a free Gemini API key → https://aistudio.google.com/app/apikey
export GEMINI_API_KEY="your_key_here"

python genai_pipeline.py

No API key? Run without AI summary:

python genai_pipeline.py --no-ai

---

Stack

Python Pandas Google Gemini API Requests Watchdog Zapier Make

---

Related project: E-Commerce Ops Dashboard — replaced 3 manual trackers with one SQL pipeline