Every feature designed to save your finance team hours of manual work
Our hybrid matching engine combines exact matching, fuzzy algorithms, and ML embeddings. Most transactions match automatically. Your team only reviews the exceptions.
Handles typos, date mismatches, formatting differences
Learns from your corrections
Processes 1M transactions in minutes
Confidence scoring for every match
Stop manually mapping columns. Our LLM detects transaction IDs, amounts, dates, descriptions—even in messy formats. No configuration needed.
Works with Payment Processors, Bank statements, Custom formats, Raw Data Formats
Detects currencies, date formats automatically
Handles multi-currency transactions
Suggests column mappings with 95%+ accuracy
Running payments, refunds, payouts, and chargebacks? Create separate Spaces for each. Keep data isolated, reports clean, and teams focused.
Unlimited spaces per account
Per-product audit trails
Role-based access per space
Independent reconciliation workflows
Every correction you make teaches the system. Historical learning means matching accuracy improves continuously without manual retraining.
Learns from your match approvals/rejections
Adapts to your business logic automatically
No ML expertise required
Accuracy improves 2-5% over first 90 days
Every match decision logged. Explainable AI shows why transactions matched. Export audit reports in one click. Built for regulated financial institutions.
Immutable audit logs
Match explanations for transparency
User action tracking
One-click audit report export
import kosha
client = kosha.Client(api_key='...')
# Upload transactions
result = client.upload_csv(
'transactions.csv'
)
# Get matches
matches = client.get_matches(
confidence_min=0.90
)
# Export results
client.export_csv(
workspace_id='payments',
format='quickbooks'
) RESTful API for custom integrations. Webhooks for real-time notifications. SDKs in Python, Node.js, Ruby. Built by developers, for developers.
Complete REST API documentation
Webhook support for events
SDKs in popular languages
Rate limits: 1000 req/min (Enterprise)
Test sandbox environment