Your data is not used to train AI models
Synota does not use customer invoices, contract data, or system data to train external AI models or shared datasets.
Automated Analysis Framework
How Synota Uses AI.
Synota uses automated analysis tools, including language models, to help interpret invoice documents and identify potential discrepancies.
These tools operate within a structured system designed for reliability and control:
AI is used to interpret and organize data
Synota performs all financial calculations and reconciliation logic
Customer data is not used to train external models
Customers remain in control and review all outputs
Capabilities & Limits
What AI Does(and Doesn’t Do)
AI helps with
Synota uses automated tools to:
- +Extract invoice data from PDFs
- +Categorize charges into standardized billing groups
- +Interpret billing structures (e.g., tariffs, taxes, regulatory charges)
- +Generate summaries of invoice and reconciliation differences
AI does not
- Perform financial calculations
- Determine settlement amounts
- Modify contracts or billing terms
- Execute transactions or system changes
All calculations and settlement logic are performed directly within Synota’s platform.
Process Architecture
How the System Works
Invoice reconciliation is not handled by a single AI request.
Instead, Synota uses a multi-step process:
- 1
Invoice data is extracted and structured
- 2
Charges are categorized and interpreted
- 3
Contract and reconciliation data are retrieved
- 4
Deterministic calculations are performed
- 5
Results are summarized for review
This approach improves accuracy, transparency, and auditability.
Accuracy & Review
Synota’s automated analysis is designed to assist invoice review, not replace it. Because utility invoices vary widely, some classifications may require validation. Users should review results alongside original invoices, contract terms, and reconciliation outputs.
Trust & Privacy
Data Privacy & Model Use.
Your data is not stored by AI systems
Customer data is processed only to generate results. All documents and outputs remain within Synota’s systems.
Reference Data & Billing Context
Synota may use structured reference data to help interpret invoice charges. These sources provide context for interpreting invoice data. All reconciliation calculations are still performed within Synota.
Reference Data & Billing Context
Synota may use structured reference data to help interpret invoice charges.
This can include:
- •utility tariffs
- •tax rules
- •regulatory charge definitions
- •other trusted industry sources
These sources provide context for interpreting invoice data. All reconciliation calculations are still performed within Synota.
Interactive Tools
Chat & Interactive Analysis
What the chat feature does
The “Ask Synota” feature allows users to:
- •ask questions about invoices and charges
- •explore reconciliation results
- •receive summaries and explanations
What the chat feature does not do
The chat feature is read-only.
It cannot:
- •modify contracts
- •change calculations
- •affect settlement outcomes
- •update system data
It is designed only to help users understand existing information.
Reliability & Testing
Synota maintains a structured testing process for automated analysis workflows.
- AI is limited to interpretation tasks
- Calculations are deterministic
- Outputs are tested against known invoice scenarios
- Results are reviewable and auditable
System Safeguards
- Automated features do not execute system changes
- Core financial logic is isolated from AI components
- Outputs are stored and traceable
- Users remain responsible for final review
Security & Compliance
Synota’s use of automated analysis tools follows the same security and data handling standards applied across the platform.
- •Customer data is handled in accordance with Synota’s security practices
- •AI-assisted features operate within Synota’s controlled infrastructure
- •External services are used only for real-time processing
For more information, see:
Synota’s automated analysis tools are designed to support—not replace—human review. Customers remain in control of how results are used and should validate outputs alongside source documents and contracts.