JMHZ and Czech payroll reporting — AI as a control assistant

Who is this guide for?

This guide is for small accounting and payroll offices, payroll specialists, and anyone preparing payroll or reporting inputs in a Czech reporting context who need to organize incomplete client information safely.

What this guide helps with

The workflow helps safely organize incomplete input data prior to professional processing:

• sort raw client inputs;
• highlight missing or ambiguous details;
• draft control questions for the client or internal team;
• separate confirmed facts from assumptions;
• list check points for verification in official sources or internal methodology.

How to use this guide

To use AI as a preparation and control assistant, follow these steps:

1. Anonymize the input: Remove all personal and sensitive details from the raw data.
2. Paste the prompt: Copy the text from the "Prompt / materials" section and paste it into the AI tool.
3. Paste the anonymized inputs: Add raw data or client notes to the chat.
4. Manually review the output: Verify all findings before proceeding with professional payroll processing.

Safe-use rules

• Complete Anonymization: Completely remove names, birth numbers, bank accounts, client names, internal IDs, and other personal or sensitive details.
• No Sensitive Documents: Never paste real payroll sheets, access credentials, or complete internal manuals into public AI tools.
• AI Does Not Decide: The AI acts strictly as a preparation and control assistant. Never treat outputs as payroll, tax, accounting, or legal advice.
• Verify Official Sources: Cross-check all ambiguities and legal requirements against official legislation, regulations, or internal methodologies.
• Specialist Responsibility: The AI does not replace a human expert. The payroll or accounting specialist remains fully responsible for the final outputs.

The sample prompt below shows the direction of the future working package. It helps prepare control questions, identify missing information and list verification points from anonymized payroll inputs. It is not payroll, accounting, tax or legal advice.

- Work only with anonymized information.
- Do not make a final professional decision.
- Do not say that a filing is correct, complete or ready to submit.
- Mark every uncertain point as “To verify”.
- If information is missing, explain why it matters.
- If an official source is needed, write “verify in official sources or internal methodology”.
- Separate confirmed information from assumptions.

WHEN I PASTE ANONYMIZED INPUTS, CREATE OUTPUT IN THIS STRUCTURE:

1. CONFIRMED INFORMATION
- only information explicitly stated in the input.

2. MISSING OR UNCLEAR INFORMATION
- missing data, conflicts and incomplete information.

3. CONTROL QUESTIONS
- questions for the client or internal team.

4. POINTS TO VERIFY
- items that must be checked in official sources or internal methodology.

5. NOTE FOR PROFESSIONAL REVIEW
- remind the reader that final assessment, processing and submission remain with the payroll/accounting specialist.

Always end with:
“This output is only a working control aid. Final professional assessment remains with the responsible human specialist.”

Example input

Here is an example of anonymized payroll details that you can use as input:

Client sent onboarding details for a new employee. The message lists the start date, contract type, and weekly hours. However, some identification details are missing, changes throughout the month are not confirmed, and it is unclear which data points are final.

Short example output

Sample format of the output prepared by the AI assistant for your review:

1. CONFIRMED INFORMATION
- Contract type and weekly hours.
- Employee's start date.

2. TO VERIFY
- Missing identification details (e.g. health insurance provider).
- Any mid-month contract changes or prior employment in the same month.

3. CONTROL QUESTIONS
- Which of the provided inputs are final and which require further validation with the client?

Note: Final assessment and submission remain with the payroll/accounting specialist.

What to check in the output

Before utilizing the output from the AI assistant, verify the following:

• Personal and Sensitive Data: Have all personal, sensitive, and identification details been thoroughly removed?
• Separation of Assumptions: Are confirmed facts clearly separated from unverified assumptions?
• Ambiguity Flags: Are all unclear details and missing inputs explicitly marked as "To Verify"?
• No AI Decisions: Does the output avoid making final payroll, accounting, tax, or legal decisions that the AI should not make?
• Official References: Do the verification points direct you to official regulations, tax laws, or internal manuals?
• Human Check: Is a final manual validation of all calculations and findings performed by a payroll specialist?

Related guides

• From client chat to clear brief – Extract structured requirements and draft clean briefs from raw client communications.
• From inquiry to next step – Evaluate incoming sales inquiries, assess project fit, and outline response scopes.