Skip to main content

How Circula AI works

Everything you need to know about Circula AI

Özge E. avatar
Written by Özge E.
Updated today

At a glance

  • Circula AI supports expense management in two core areas:

    • Intelligent extraction and completion of expense data

    • Early detection of common document and compliance issues

  • The goal is simple: increase the share of expenses that require little to no manual input across the entire workflow.

  • AI supports existing processes. Approvals, checks, and final responsibility always remain with the company.


Overview

Circula uses artificial intelligence to make expense management more accurate, efficient, and scalable.

Circula AI focuses on:

  • Extracting and completing expense data with very high accuracy

  • Validating documents against common issues and legal requirements

By combining highly accurate data extraction with reliable validation, Circula helps ensure that expenses are submitted correctly from the start and move smoothly through submission, approval, and controlling.


Data extraction and completion

AI-powered receipt recognition

When uploading a receipt, Circula AI automatically recognizes and pre-fills key information, including:

  • Date

  • Amount and currency

  • Merchant details

  • VAT rates and amounts

  • Invoice or receipt number

  • Merchant IBAN

  • Due date

  • Service date

Circula uses modern Large Language Models (LLMs) to extract this data. Compared to traditional OCR approaches, this leads to significantly higher accuracy and more stable results across different receipt layouts, languages, and image qualities.

This means fewer incorrect fields and fewer manual corrections.

Employees can always review and adjust recognized data before submitting an expense.


Automatic multi VAT split

If a receipt contains multiple VAT rates, Circula automatically splits the expense across the detected VAT rates during the controlling process, where the system is confident in the accuracy.

This reduces manual VAT adjustments for accountants and makes the controlling process faster and more straightforward.


Automatic category prediction

After uploading a receipt, Circula AI suggests the most likely expense category.

The model learns from historical expense patterns within the organization, which improves suggestion quality over time.

The suggested category can always be reviewed and adjusted before submission.


Compliance and validation

In addition to completing data, Circula AI helps identify common issues early in the workflow.

Duplicate detection

Circula helps identify potential duplicate expenses.

Receipts are compared based on key fields such as invoice date, amount, time, and invoice number. If these match, a warning is shown to Root Accountants.

This helps prevent double reimbursement and strengthens process transparency.

Tip

  • More information about duplicate detection can be found here.


Unreadable receipt detection

If a receipt is blurry, too dark, or otherwise difficult to read, Circula displays an immediate warning and asks the user to upload a clearer image.

Addressing quality issues early helps avoid delays later in the process.

Tip

  • More information about unreadable receipt detection can be found here.


Transaction receipt detection

Circula detects transaction confirmations and checks documents during expense submission.

If a document does not meet VAT invoice requirements, it is flagged directly. This helps reduce document-related rejections and improves overall submission quality.

Tip

  • More information about transaction receipt detection can be found here.


Company address validation (Beta)

For invoices above €250, Circula validates whether the required company name and address are included, in line with German VAT law (UStG).

If required information is missing or incorrect, the employee is informed immediately. This helps prevent repeated back-and-forth between employees and finance teams.

This feature is currently in Beta.


AI and data privacy

Circula processes data strictly for the purpose of extracting, enriching, and validating expense information.

The following data categories may be processed during AI-supported analysis:

  • Extracted receipt data such as amount, date, merchant, VAT, invoice number, and currency

  • Customer configuration data such as categories, cost centers, and policy rules

  • Limited contextual metadata from recent expense entries

  • Receipt or invoice files for temporary analysis

  • AI-generated suggestions such as category proposals or validation flags

Data handling principles

  • Processing is EU-only

  • Customer data is never used to train AI models

  • Data processing is request-bound and transient

  • AI-related data retention is limited to 24 hours for performance optimization

Note

  • For detailed information, please refer to our Data Processing Agreement (DPA).


Impact on the workflow

With accurate data extraction and early validation:

  • Employees spend less time entering and correcting data

  • Supervisors receive more complete and consistent submissions

  • Accountants spend less time on manual adjustments

As a result, more expenses can move through the workflow with little to no manual input.


Requirements and notes

  • Available AI features may vary depending on configuration and usage.

  • AI supports existing processes but does not replace them.

  • Approvals, checks, and responsibility remain with the company.


Conclusion

Circula AI is designed to reduce manual effort while maintaining transparency and control.

By combining highly accurate data extraction with reliable validation, Circula helps ensure that expenses are submitted correctly and processed more efficiently, with significantly less manual intervention.


Do you still have questions?

💬 Chat Support
📞 +49 30 588 491 01
✉️ support@circula.com

Did this answer your question?