Ukrainian compliance — AI Tools

AI-Assisted Accounting

Rule-based anomaly detection (4 types). Posting suggestions from historical patterns. OCR interface ready for GPT-4 Vision, Google Vision or Tesseract — built on the Ports & Adapters pattern for easy provider swap.

4 anomaly types Posting suggestions OCR: GPT-4 Vision ready Ports & Adapters pattern

Four rule-based invoice anomaly types

The detector scans invoices over a configurable time window and returns anomalies with severity scores. The approach is deliberately rule-based — the rules are auditable, explainable, and require no training data.

high Large invoice

Amount > mean + 3σ per customer (last 90 days)

high Duplicate invoice

Same amount + customer within 7 days

medium Off-hours posting

Outside 08:00–20:00 Kyiv or weekend

low Round amount

Total divisible by configurable threshold

Scan window configurable per use case — e.g. last 24h for daily review, last 72h for weekend catch-up
Each anomaly: type, invoice ID, description, severity (low / medium / high)

Posting suggestions from historical patterns

The suggestion engine learns from the last 90 days of posted journal entries. When an accountant creates a manual entry, the engine returns the most frequently used debit/credit account pairs — ranked by confidence.

Three ways to look up suggestions

By counterparty — top account pairs for this supplier or customer (last 90 days)
By similar amount — entries within ±20% of the entered amount
By document type — keyword matching on entry description for known document types
Global top-N — most commonly used account pairs across the entire company

Each suggestion: Confidence score, UsageCount (how many times used), Description
Higher UsageCount → higher confidence — improves automatically as more entries are posted

Document recognition — ready for GPT-4 Vision and Google Vision

The OCR layer is built on the Ports & Adapters pattern. The interface defines the contract — any vision model can be plugged in without changing business logic. Today the mock is active; a real provider replaces it by swapping one dependency.

OpenAI GPT-4 Vision Google Cloud Vision Tesseract (local) MockOCRProvider ← active now

OCRProvider interface: one Scan() method, any implementation — provider swap without business logic change
DocumentScanResult: DocumentType, ЄДРПОУ, amounts, RawText — structured output from any provider
OCRService wraps provider + posting suggestions — recognised document immediately gets account hints
Ready-to-plug: OpenAI GPT-4 Vision · Google Cloud Vision · Tesseract (local, open-source)

How the three tools work together in practice

Anomaly detection, posting suggestions and document recognition are independent components that complement each other. A typical daily workflow:

Morning: scan anomalies for the last 24 hours

Review flagged invoices by severity. High-severity items first — duplicates and outlier amounts resolved before the day's posting begins.

During the day: create manual entries with suggestions

The system shows top-3 suggested debit/credit pairs. One-click accept fills the account fields. If wrong — override, and the next suggestion improves.

When a paper document arrives: photograph → automatic recognition

Upload a photo or scan. Once a real OCR provider is connected, the system extracts supplier, ЄДРПОУ, date, amount, VAT — and immediately offers posting suggestions based on this counterparty's history.