🧠

Explainable AI

Each recommendation includes an easy explanation of the inputs and reasoning used.

πŸ”

Transparent Process

We document models, thresholds and updates so you can see how the system evolves.

πŸ›‘οΈ

Ethical Guardrails

Built with fairness testing, conservative defaults and user-first privacy choices.

🧩 Model Versions & Lifecycles

Current components, versions and the latest changes.

Component Version Purpose Last Update Change
Financial Analysis Engine v1.7.3 Spending patterns & budget insights 2025-10-20 Reduced false spikes; seasonal spend modelling
Risk Assessment v0.9.9-beta Debt/overdraft early-warning 2025-10-12 Threshold tuning; user adjustable sensitivity
NLP Assistant v2.1.0 Natural language entry & explanations (TR/EN) 2025-09-28 Better date/money parsing; multi-language fixes

Note: Some version bumps are delivered via remote config (no app update). Critical changes trigger an in-app notice.

πŸ”¬ Our AI Models

πŸ“Š

Financial Analysis Engine

Machine Learning

Analyzes spending, income trends and behavior to drive insights and budgets.

  • Pattern recognition on financial time series
  • Anomaly detection for unusual spend
  • Budget optimization heuristics
  • Cash-flow prediction
🎯

Investment Predictor

Deep Learning

Reads market trends and risk tolerance to suggest portfolio adjustments.

  • Trend analysis & regime detection
  • Portfolio optimization (constraints aware)
  • Risk profiling
  • Forward-looking scenarios
⚑

Risk Assessment

Ensemble Methods

Combines signals to surface debt and overdraft risks early.

  • Credit/obligation pressure
  • Volatility & burn-rate stress
  • Fraud/abuse heuristics
  • Scenario stress tests
πŸ’¬

Natural Language Processor

Transformer Models

Understands your free-text (or voice) and produces structured entries.

  • Financial intent & slot extraction
  • Context-aware clarifications
  • Turkish & English
  • Lightweight sentiment

πŸ”„ How AI Makes Decisions

A conservative pipeline with human-readable explanations at the end.

1

Data Collection

User-entered records; optional telemetry (no raw PII).

2

Pattern Analysis

Time-series features & seasonality.

3

Risk Screening

Debt, overdraft, recurring drains.

4

Personalization

Budgets/goals context and preferences.

5

Validation

Cross-checks & thresholds.

6

Explanation

Plain-language rationale + edit/ignore.

πŸ“¦ Data Practices & Retention

We process

User-entered finance data (income, expense, goals), app settings, optional anonymous metrics.

We don’t

Store card numbers, full IBAN or raw passwords. External bank logins are not required.

Data Type Purpose Retention Deletion
Financial Records Core features Until user deletes In-app: Settings β†’ Data β†’ Delete My Data
Error/Crash (optional) Diagnostics 90 days Consent off + early deletion on request
Usage Analytics (optional) Product improvement 180 days Opt-out stops future; past anonymized in 30 days

Portability: Export CSV/JSON via Settings β†’ Export.

πŸ“‘ Telemetry Matrix

Event Fields PII? Purpose
app_open app_version, platform, locale No Compatibility & localization
txn_add amount_range, category, currency No (range) UX improvement & suggestions
crash stacktrace, device_model Optional (user_id with consent) Bug fixing
πŸ‡ͺπŸ‡Ί
GDPR
Aligned
πŸ›‘οΈ
ISO 27001
Aligned
🏦
PCI DSS
Aligned
πŸ‡ΊπŸ‡Έ
SOC 2
Aligned

πŸ›  Status & Incident Log

Events in the last 12 months affecting user data would appear here.

No known incidents impacting user data in the last 12 months.

πŸ—’ AI Changelog

2025-10-20 – Analysis v1.7.3

Reduced weekly-spend false positives from 18% β†’ ~7% on validation sets.

2025-10-12 – Risk v0.9.9-beta

Added user sensitivity control; tuned thresholds for overdraft warnings.

πŸ“š Research & References

  • Thaler & Sunstein – Nudge: budgeting & behavior framing
  • Survey: Anomaly detection in financial time series
  • Privacy patterns for mobile analytics (event minimization)

Whitepaper coming soon: FinFin AI Technical Whitepaper (PDF)

πŸ’¬ Questions About Our AI?

We believe in complete transparency. Ask us anything.

Contact AI Ethics Team Read Our AI Whitepaper