Upload your bank statements. FinTrack uses AI to extract every transaction,categorize spending, detect anomalies, and answer any question about your finances —no bank credentials required.
Deutsche BankPostbankSparkasseCommerzbankHypoVereinsbankRaiffeisenbankVolksbankDeutsche BankINGINGDKBRevolutComdirectConsorsbankN26Vivid Money TomorrowDeutsche BankPostbankSparkasseCommerzbankHypoVereinsbankRaiffeisenbankVolksbankDeutsche BankINGINGDKBRevolutComdirectConsorsbankN26Vivid Money Tomorrow
98%+categorization accuracy
17+Banks supported in germany
MultipleAI engines working together
NObank credentials needed
How it works
Four steps from PDF to insight.
1UploadDrop your monthly bank PDF — the same one your bank emails you.
2ExtractGemini Vision parses every transaction from the raw PDF in seconds.
3CategorizeRules handle known merchants instantly. AI covers the rest with 98%+ accuracy.
4AnalyzeExplore charts, set budgets, detect anomalies, and chat with your data.
Features
Everything you need. Nothing you don't.
PDF UploadDrop any German bank statement. Gemini Vision extracts every transaction in a single API call.
98%+ AI AccuracyHybrid rule engine + Gemini fallback. Every correction you make trains a rule — the AI is needed less over time.
Anomaly DetectionZ-score analysis flags unusual transactions automatically — like a €41 restaurant bill when you normally spend €12.
Natural Language QueriesAsk "How much did I spend on food in October?" — the engine generates SQL from plain text and returns a direct answer.
AI Chat AssistantA conversational advisor with full access to your transaction history. Answers follow up questions in context.
Budget TrackingSet monthly limits per category. Visual progress bars turn amber and red as you approach or exceed your budget.
Spending AnalyticsMonthly bar charts, category donut breakdowns, and 6-month trend views — all from your uploaded statements.
No Bank Access RequiredWorks entirely from the PDF you already receive. No credentials, no open banking, no third-party connections.
The AI story
Not one AI. Four working together.
Each layer is specialized. The rule cache handles what it knows. Gemini handles the unknown. Nothing is over-engineered — every AI call earns its cost.
01
PDF Extraction
Gemini Vision
One API call processes an entire bank statement. Gemini Vision reads the raw PDF and returns structured transaction data — date, amount, counterpart, description.
02
Categorization
Rule Cache + LLM Fallback
Known merchants are matched instantly by a rule cache that learns from every correction you make. Unknown transactions fall through to Gemini. Over time, fewer AI calls are needed.
03
NL Query Engine
Text → SQL → Answer
A plain-language question becomes a SQL query via Gemini, runs against your real transaction data, and returns a precise answer — not a vague summary.
04
Chat Assistant
Conversational Finance Advisor
Full transaction context is injected into every chat session. Ask follow-up questions, compare months, get advice — all grounded in your actual spending data.
Tech stack
Spring Boot 3PostgreSQLGemini 2.5 FlashNext.jsDocker
Your finances. Finally clear.
No subscription. No bank access. No setup beyond uploading a PDF. Try the demo account to explore with real-looking data.