Bank Statment Analyser
An intelligent bank statement analysis tool that leverages artificial intelligence to automatically process, categorise, and analyse financial transactions from bank statements. The app uniquely shows users their redacted document before anything is processed, proving their sensitive information is protected. Combined with a cloud LLM (Ollama)
Technologies Used
Python, FastAPI, Vue.js, API development, AI, Celery
The Goal:
IntelliBank transforms personal financial management by allowing users to upload personal bank statement PDFs, automatically redact sensitive PII before any processing, receive AI-generated financial insights from a local LLM, and visualise those insights through an interactive dashboard — all without their raw personal data ever leaving their control. The platform automatically categorises transactions, identifies spending patterns, detects financial anomalies, and provides personalised recommendations to improve users' financial health.
1
The Challenge:
PII Accuracy: Balancing complete detection across diverse bank layouts while minimising false positives.
Performance: Client-side redaction and managing large 20MB PDFs.
Privacy & Security: Guaranteeing unredacted data never leaves the browser and sanitising inputs for Ollama.
LLM Integration: Extracting structured, accurate financial data from fragmented, redacted text reliably.
UX Complexity: Building intuitive manual correction interfaces and communicating redaction limits clearly.
2
The Result
Privacy-First AI: Secure financial analysis without exposing raw sensitive data to third-party services.
Deep Insights: Interactive dashboards for spending categories, recurring transactions, and fraud alerts.
Robust Processing: Dual-layer redaction with exports to JSON, Excel, and styled PDF reports.
User Empowerment: Streamlined upload workflows, tag management, and multi-statement trend views.
Multi-Sector Value: Tailored reporting for individuals, small businesses, and professional accountants.
3
Other Projects
