The Finance Tracker's automation architecture is as follows:
Aim: Build a basic script that checks emails for bank notification and organises my important bank transactions.
Fundamentals: Filter emails for transaction details and organise them
Modules/Stages:
Part One: Core Engine
- Parsing texts → structured transaction objects
- Categorising transactions
- Storing/organising transactions
Part Two: Insights Engine
- Aggregations (daily/weekly/monthly totals)
- Category breakdowns
Part Three: Visualization
- Dashboard
Part Four: Data ingestion
- Email connection / extraction
Optional: Intelligence layer
- LLM-generated explanations/advice
Additionals:
- Dashboard to visualise spending
- AI to give advice (or better still, generate a prompt that includes a summay of my spending and demands brief questioning to understand one's spending and give a resonable actionable goal)
Possible tools:
- Python streamlit
Connections/Integrations:
- Email reading/filtering
- Banks being used / phrases to identify transaction details
- Your personal LLM to give financial advice / a chatgpt tool ot service that provides that feature