Delivered Submodules & Capabilities
In depth breakdown of the 3 native components forming the Syna AI Copilot & Automation workspace:
Document OCR & extraction
High-accuracy optical character recognition (OCR) and document intelligence designed to read scanned physical receipts, utility bills, and vendor PDF invoices. Automatically extracts supplier names, TINs, invoice dates, line items, and tax amounts.
- ✓Automatic extraction of vendor TIN, invoice serial, and line totals
- ✓Handling of physical mobile photos, scans, and digital PDFs
- ✓Automatic matching of extracted vendors to existing master records
- ✓side by side visual document verification interface
Transaction proposal engine
Analyzes extracted document data and historical accounting entries to recommend appropriate debit and credit accounts, tax codes, and cost centers. Never posts autonomously without human approval.
- ✓Intelligent chart of accounts coding recommendations
- ✓Automatic 18% VAT and 2.5% SSCL split proposals
- ✓Learns from accountant corrections to improve future recommendations
- ✓Complete human in the loop verification gate before posting
Natural language financial queries
Allows business founders and finance managers to query financial data in plain English or Sinhala (e.g., 'What was our total VAT liability for last month?' or 'Which 5 customers have overdue balances over 60 days?').
- ✓Natural language querying over ledger, AR, and AP records
- ✓Instant generation of ad-hoc financial tables and summary metrics
- ✓Short-term cash flow projections based on open receivables and bills
- ✓Strict data isolation ensuring customer financial records remain private
Core Capabilities & Accounting Logic
human in the loop Financial Safety
AI must never post financial transactions without verification. Syna operates as a tireless assistant that prepares proposals for qualified accountants to approve.
- Presents side by side visual document verification before posting
- Highlights low-confidence fields with visual warning indicators
- All AI proposals are fully audited with human confirmation timestamps
Sri Lanka Tax Context Awareness
Trained to understand local tax terminology, IRD invoice formats, and the distinction between standard rated, exempt, and SSCL-liable supplies.
- Automatically separates 18% VAT and 2.5% SSCL during receipt extraction
- Recognizes Inland Revenue Department TIN and VAT registration numbers
- Detects duplicate receipts to prevent double-claiming of input tax
Sri Lanka Statutory & Accounting Standards
Admissibility of electronic records and scanned documentation for audit.
Ensures financial records are processed securely without unauthorized AI training.
Standard Reports Generated
| Report Name | Frequency | Description |
|---|---|---|
| AI Extraction Accuracy Audit | Monthly | Review of extraction confidence, user overrides, and continuous learning. |
Why Financial AI Must Never Post Autonomously
Autonomous AI agents that execute financial transactions without human verification create major fiduciary risks. If an algorithm misinterprets a handwritten receipt or miscategorizes capital expenditure as operational expenses, errors compound throughout the general ledger.
SyncBooks guarantees audit protection by keeping humans in the loop. Syna handles the tedious work of reading receipts and proposing accounts, while authorized accountants retain complete control.
Frequently Asked Questions
Is our financial data private when using Syna AI?↓
Yes. All prompts and documents are isolated in your encrypted tenant environment. Your company records are never used to train public foundation models.