Skip to content
D
GoBD-compliant §203 StGB-compliant Q1

Entertainment Expense Agent

Verify entertainment receipts, calculate deductibility, handle VAT correctly.

Recognises entertainment receipts, verifies mandatory fields per §4 Abs. 5 Nr. 2 EStG, calculates the 70% deductibility and checks input tax deduction.

Analyse your process
Airbus Volkswagen Shell Renault Evonik Vattenfall Philips KPMG

Rule-based receipt-field checks per Paragraph 4 EStG, LLM plausibility of business reason

The agent validates mandatory fields per Paragraph 4(5)(2) EStG deterministically, calculates the 70 percent deductibility by rules, and uses LLM assessment only to plausibility-check the business reason from the free-text field.

Outcome: Throughput per entertainment receipt reduced from 8 to under 2 minutes, input-VAT certainty on 100 percent of mandatory fields, and no auditor queries on formalities.

67% Rules Engine
22% AI Agent
11% Human

The clean split between rule and AI protects input-VAT deduction and scales at the same time:

480 euros of entertainment, denied over a missing attendee line

Entertainment receipts do not fail tax audits because of incorrect amounts. They fail because of missing mandatory fields - a formality error that destroys the entire business expense deduction. With EUR 10.9 billion (USD 12 billion) in additional tax assessments from audits in 2024 (source: Federal Ministry of Finance, November 2025), entertainment expenses are among the items auditors systematically target. Every receipt without complete information is a hit.

Form errors wipe out the entire deduction

The arithmetic is simple: 70 percent deductible, 30 percent not. Nobody fails at that. The problem lies in the five mandatory fields required under German tax law: location, date, attendees, business purpose and amount. If one is missing, the tax authority does not strike 30 percent - it strikes everything. Location and date appear on the restaurant receipt. Attendees and purpose are where it gets critical. In practice, the most common gap is an incomplete attendee list or a business purpose so vague - “business dinner” without further detail - that the tax authority does not accept it. (US: The IRS requires similar substantiation under IRC Section 274; (UK:) HMRC applies the “wholly and exclusively” test for business entertainment.)

A concrete scenario: a sales director entertains four clients after a product presentation. The EUR 480 (USD 530) receipt is correct, the restaurant included all information on the till receipt. But the entertainment form lists only “client meeting” as the purpose and one attendee name is missing. During a tax audit three years later, the full EUR 480 is disallowed - not because the amount was unreasonable, but because two lines were incomplete.

Tightened documentation requirements raise the bar further

Updated guidance from the German tax authority in November 2025 tightened the documentation rules for entertainment expenses. Restaurants with electronic point-of-sale systems must produce machine-generated receipts secured by a certified technical security device. Handwritten receipts or simple printouts without security markings are no longer accepted. For companies, this means: even a substantively correct entertainment receipt can fail if the underlying restaurant receipt does not meet the formal requirements. A compliant receipt is recognisable by its transaction number, the till system’s serial number or a printed QR code.

Nine decision steps between receipt and posting

The Decision Layer breaks entertainment receipt verification into nine steps with clear assignment: who decides - rule engine, AI, or human?

Receipt classification uses the LLM to recognise whether the document is an entertainment receipt at all. Then the rule engine checks the five mandatory fields and the completeness of the attendee list deterministically. These are binary questions - present or not. There is no room for discretion and no reason for human intervention.

The plausibility of the business purpose is assessed by the LLM. “Project meeting - logistics digitalisation with Company X” is plausible. “Dinner” is not. The model recognises patterns and prompts the submitter to add detail before the receipt reaches accounting.

The deductibility split (70/30), input tax deduction and account coding run rule-based. Audit-compliant archiving with a timestamp closes the process. All nine steps are documented and traceable for a subsequent tax audit.

Reasonableness remains a human decision

One decision step is deliberately handled by no algorithm: the reasonableness check. Whether EUR 120 (USD 132) per person at a business dinner with three clients is reasonable depends on industry, context and the business relationship. A dinner during contract negotiations follows different standards than a lunch after an introductory meeting. The agent gives the decision-maker the facts - amount per person, comparison with historical values, ratio to the business volume involved. The approval or escalation is the human’s call.

This boundary is not a technical limitation. It is governance. An entertainment expense agent that independently judges reasonableness would automate discretionary decisions that a tax auditor would challenge. Responsibility stays where it belongs.

Micro-Decision Table

Who decides in this agent?

9 decision steps, split by decider

67%(6/9)
Rules Engine
deterministic
22%(2/9)
AI Agent
model-based with confidence
11%(1/9)
Human
explicitly assigned
Human
Rules Engine
AI Agent
Each row is a decision. Expand to see the decision record and whether it can be challenged.
Recognise entertainment receipt Is this receipt an entertainment receipt? AI Agent Employee

LLM classification of receipt type

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Challengeable by: Employee

Verify mandatory fields Are location, date, attendees, purpose and amount present? Rules Engine Employee

Deterministic check against Paragraph 4 Abs. 5 Nr. 2 EStG

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Challengeable by: Employee

Attendee list complete Are all entertained persons listed by name? Rules Engine Employee

Deterministic check - statutory requirement

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Challengeable by: Employee

Business purpose plausible Is the stated business purpose credible? AI Agent Employee

LLM assesses free-text justification for plausibility

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Challengeable by: Employee

Calculate deductibility What is the deductible portion? Rules Engine Auditor

70% of net amount per EStG

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Challengeable by: Auditor

Check reasonableness Is the entertainment amount reasonable? Human Auditor

Human judgement for borderline cases and unusually high amounts

Decision Record

Decider ID and role
Decision rationale
Timestamp and context

Challengeable: Yes - via manager, works council, or formal objection process.

Challengeable by: Auditor

Check input tax deduction Is input tax deduction per Paragraph 15 UStG permitted? Rules Engine Auditor

UStG-compliant check of receipt requirements

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Challengeable by: Auditor

Account coding Which journal entry is created? Rules Engine Auditor

Posting logic: 70% deductible, 30% non-deductible, input tax

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Challengeable by: Auditor

GoBD archiving Is the receipt archived GoBD-compliantly? Rules Engine Auditor

Automatic archiving with timestamp and link to posting

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Challengeable by: Auditor

Decision Record and Right to Challenge

Every decision this agent makes or prepares is documented in a complete decision record. Affected parties (employees, suppliers, auditors) can review, understand, and challenge every individual decision.

Which rule in which version was applied?
What data was the decision based on?
Who (human, rules engine, or AI) decided - and why?
How can the affected person file an objection?
How the Decision Layer enforces this architecturally →

Does this agent fit your process?

We analyse your specific finance process and show how this agent fits into your system landscape. 30 minutes, no preparation needed.

Analyse your process

Governance Notes

GoBD-compliant §203 StGB-compliant

GoBD relevance: high - entertainment receipts are a frequent focus during tax audits. Paragraph 4 Abs. 5 Nr. 2 EStG defines the requirements: mandatory fields, attendee list, business purpose. The 70% deduction rule is deterministic. The reasonableness check requires human judgement because the tax principle of proportionality applies. Input tax deduction per Paragraph 15 UStG is checked separately.

§203 StGB-relevant data is encrypted end-to-end and never passed to AI models in plain text.

Process Documentation Contribution

The Entertainment Expense Agent documents: the receipt classification, all verified mandatory fields, the plausibility assessment of the business purpose, the deductibility calculation (70/30 split), the reasonableness check and the input tax deduction. During a tax audit, every entertainment receipt is fully traceable.

Assessment

Agent Readiness 82-89%
Governance Complexity 21-28%
Economic Impact 68-75%
Lighthouse Effect 18-25%
Implementation Complexity 24-31%
Transaction Volume Weekly

Prerequisites

  • Entertainment receipt template with mandatory fields
  • Configured reasonableness thresholds
  • ERP system with separate accounts for deductible/non-deductible
  • GoBD-compliant archiving system

Infrastructure Contribution

The Entertainment Expense Agent uses the receipt classification of the Travel Expense Agent and the posting logic of the Account Coding Agent. The mandatory fields check (EStG Paragraph 4 Abs. 5) becomes the standard pattern for all tax-sensitive receipts. The 70/30 split logic is reused across expense accounting.

What this assessment contains: 9 slides for your leadership team

Personalised with your numbers. Generated in 2 minutes directly in your browser. No upload, no login.

  1. 1

    Title slide - Process name, decision points, automation potential

  2. 2

    Executive summary - FTE freed, cost per transaction before/after, break-even date, cost of waiting

  3. 3

    Current state - Transaction volume, error costs, growth scenario with FTE comparison

  4. 4

    Solution architecture - Human - rules engine - AI agent with specific decision points

  5. 5

    Governance - EU AI Act, GoBD/statutory, audit trail - with traffic light status

  6. 6

    Risk analysis - 5 risks with likelihood, impact and mitigation

  7. 7

    Roadmap - 3-phase plan with concrete calendar dates and Go/No-Go

  8. 8

    Business case - 3-scenario comparison (do nothing/hire/automate) plus 3×3 sensitivity matrix

  9. 9

    Discussion proposal - Concrete next steps with timeline and responsibilities

Includes: 3-scenario comparison

Do nothing vs. new hire vs. automation - with your salary level, your error rate and your growth plan. The one slide your CFO wants to see first.

Show calculation methodology

Hourly rate: Annual salary (your input) × 1.3 employer burden ÷ 1,720 annual work hours

Savings: Transactions × 12 × automation rate × minutes/transaction × hourly rate × economic factor

Quality ROI: Error reduction × transactions × 12 × EUR 260/error (APQC Open Standards Benchmarking)

FTE: Saved hours ÷ 1,720 annual work hours

Break-Even: Benchmark investment ÷ monthly combined savings (efficiency + quality)

New hire: Annual salary × 1.3 + EUR 12,000 recruiting per FTE

All data stays in your browser. Nothing is transmitted to any server.

Entertainment Expense Agent

Initial assessment for your leadership team

A thorough initial assessment in 2 minutes - with your numbers, your risk profile and industry benchmarks. No vendor logo, no sales pitch.

30K120K
1%15%

All data stays in your browser. Nothing is transmitted.

Frequently Asked Questions

What happens when mandatory fields are missing?

The agent documents exactly which mandatory fields are missing and routes the receipt to the submitter. Missing attendees, missing purpose or missing location means the entire receipt is non-deductible. The agent flags this before creating the posting.

How is reasonableness assessed?

The agent checks the amount against configurable thresholds (per person, per occasion). Standard cases pass automatically. On breach, it is routed to the supervisor or accounting. The final assessment remains with the human because reasonableness is a discretionary decision.

Why is the plausibility check AI-assisted?

The business purpose is a free-text field. Checking whether a purpose like project meeting or customer acquisition sounds plausible requires language understanding. The AI Agent assesses plausibility - at low confidence, it escalates to a human.

What Happens Next?

1

30 minutes

Initial call

We analyse your process and identify the optimal starting point.

2

1 week

Discover

Mapping your decision logic. Rule sets documented, Decision Layer designed.

3

3-4 weeks

Build

Production agent in your infrastructure. Governance, audit trail, cert-ready from day 1.

4

12-18 months

Self-sufficient

Full access to source code, prompts and rule versions. No vendor lock-in.

Implement This Agent?

We assess your finance process landscape and show how this agent fits your infrastructure.