Skip to content
W K
GoBD-compliant §203 StGB-compliant Q2

Provisions Agent

Calculate, value and post provisions - from leave to litigation costs.

Identifies provision types, calculates leave and bonus provisions deterministically.

Analyse your process
Airbus Volkswagen Shell Renault Evonik Vattenfall Philips KPMG

Rule-based vacation and bonus provisions, LLM analysis for warranty, judgement stays with humans

The agent calculates vacation provisions and bonuses fully deterministically from collective-agreement and contract data, uses LLM analysis for warranty cases, and hands litigation provisions to legal.

Outcome: Provision calculation reduced from 3 days to 4 hours per close, 100 percent traceable basis per provision, and an audit-proof provisions schedule.

49% Rules Engine
13% AI Agent
38% Human

The 8 steps show where provisions are calculable and where they require judgement:

Provisions in an Excel template, three logics, one result

Provisions are the perennial focus of every financial audit because recognition and measurement rest on assumptions and estimates. The Provisions Agent cleanly separates what can be calculated from what requires human judgement - and makes both parts audit-proof.

Provisions Are the Perennial Focus of Every Financial Audit

No other balance sheet item attracts such regular scrutiny from statutory and tax auditors. Professional accounting bodies describe provisions explicitly as a “perennial topic” of audit season, because both recognition and measurement under commercial law and IAS 37 rely on assumptions that management must make. Enforcement authorities have repeatedly placed asset impairments and related valuation questions among their national audit priorities.

The reason is structural. A leave provision follows a clear formula. A litigation cost provision depends on the legal assessment of success probability. A warranty provision rests on historical claims experience. Three provisions, three entirely different origination logics - and in many Finance departments they still end up in the same spreadsheet template, manually carried forward each year. The statutory auditor sees a result at the end, but not which portion was calculation and which was judgement.

Leave and Bonus Provisions Can Be Formed Fully Deterministically

The Decision Layer breaks provision formation into eight steps. Three are pure calculation rules that require no estimation. The leave provision derives from outstanding leave days multiplied by the average daily rate including employer social insurance contributions. The bonus provision follows the contractual basis and the current earnings forecast. The journal entry derives deterministically from the provision type.

For a mid-sized industrial company with 800 employees that previously spent three working days in January transferring leave data from HR time tracking into a spreadsheet, the agent takes over this run monthly. Data comes directly from the ERP and personnel master data. The result is not an estimate but a traceable per-employee calculation. For the statutory auditor, the question is no longer whether the amount is plausible but only whether the calculation inputs are correct. This shortens the audit from sample testing with follow-up queries to a pure methodology check.

Warranty and Litigation Costs Remain Human Decisions

Two steps in the pyramid are deliberately not automated. Warranty provisions rest on historical claims experience and company-specific judgement that no rule engine can fully replicate. Litigation cost provisions depend on legal assessments of success probabilities that require evaluation by in-house counsel or external law firms.

Here the agent delivers not the result but the decision basis. For warranties, it aggregates claims and goodwill data from the past three years, calculates rates by product group and provides the foundation for the rate estimate. For litigation, it pulls claim amounts, prior legal costs and court instance from the legal department and hands the template to the human decision-maker. The final rate and the final provision amount stay with the human - but they are made from a structured data set, not from memory.

The Decision Layer Documents Every Valuation Audit-Ready

What matters for the statutory auditor is not whether a provision was calculated or estimated. What matters is whether the basis is documented. The Provisions Agent records per provision: type, calculation method, input data, result, the label deterministic or judgement-based, prior-year comparison and rationale for every deviation. At year-end, this produces a provisions schedule that can be submitted as an annex to the annual financial statements.

Notes disclosures under commercial law or IAS 37 are prepared as LLM drafts. This does not mean the accountant accepts them unchanged. They review whether the wording fits the valuation and approve. For Finance teams that need every hour twice during closing season, the work shifts from writing to reviewing - and above all away from the recurring auditor question of how a particular figure was derived.

Micro-Decision Table

Who decides in this agent?

8 decision steps, split by decider

49%(4/8)
Rules Engine
deterministic
13%(1/8)
AI Agent
model-based with confidence
38%(3/8)
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.
Identify provision type What type of provision is this? Rules Engine

Categorisation (leave, bonus, warranty, litigation, restoration)

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.

Calculate leave provision What is the leave provision per employee? Rules Engine Auditor

Remaining entitlement times daily rate - fully deterministic

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

Calculate bonus provision What is the bonus provision? Rules Engine Auditor

Contractual terms plus current result - deterministic

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

Estimate warranty provision How high is the warranty risk? Human Auditor

Experience, current claims rates, commercial judgement

Decision Record

Decider ID and role
Decision rationale
Timestamp and context

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

Challengeable by: Auditor

Assess litigation cost provision How high is the litigation cost risk? Human Auditor

Legal assessment of likelihood of success and damages amount

Decision Record

Decider ID and role
Decision rationale
Timestamp and context

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

Challengeable by: Auditor

Review prior-period provisions Should existing provisions be released, adjusted or retained? Human Auditor

Unchanged facts rule-based (R), revaluation human (H)

Decision Record

Decider ID and role
Decision rationale
Timestamp and context

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

Challengeable by: Auditor

Create journal entry What are the journal entries for formation, release and adjustment? Rules Engine Auditor

Posting logic: expense to provisions, provisions to income

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

Draft notes disclosures Which notes disclosures are required for provisions? AI Agent Auditor

LLM draft of notes disclosures based on provision data

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: 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

Mixed decision distribution (2H / 4R / 1A / 1 R/H). Human decisions for warranty and litigation cost provisions - both require judgement and are balance-sheet-relevant. HGB Paragraph 249 (obligation to provide), HGB Paragraph 253 (valuation) and EStG Paragraph 6 Abs. 1 Nr. 3a (tax valuation) as direct legal bases.

GoBD-compliant: every provision is documented with calculation basis, method applied and result. Discretionary decisions (warranty, litigation) are archived with the clerk's rationale. During tax audits, provisions are a frequent focus, especially the appropriateness of valuations. Paragraph 203 StGB relevant: provisions can reveal litigation risks and internal conflicts.

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

Process Documentation Contribution

Per provision type: calculation basis, method applied (deterministic or estimate), input values, result. For discretionary decisions: clerk's rationale, experience values used, comparison with prior year. For prior-period review: status (retained, adjusted, released) with rationale. Notes disclosure drafts as working papers for the close.

Assessment

Agent Readiness 66-73%
Governance Complexity 34-41%
Economic Impact 64-71%
Lighthouse Effect 24-31%
Implementation Complexity 36-43%
Transaction Volume Monthly

Prerequisites

  • ERP system with provisions module or separate provisions tool
  • Personnel master data for leave provisions (remaining leave, daily rate)
  • Contract data for bonuses (basis, percentage)
  • Legal assessments for litigation provisions
  • Historical warranty data for estimation basis

Infrastructure Contribution

The Provisions Agent establishes the pattern for hybrid agents: calculable parts fully automated, valuation decisions with the human. This pattern is reused by the Revenue Recognition Agent, Journal Entry Agent and other agents with discretionary components. The prior-period review (retain vs. adjust vs. release) is a reusable pattern for all balance sheet items that require regular review. Builds Decision Logging and Audit Trail used by the Decision Layer for traceability and challengeability of every decision.

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.

Provisions 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

Can the agent also calculate IFRS provisions per IAS 37?

Yes. The agent supports both HGB Paragraph 249 and IAS 37. The difference lies in valuation (HGB: most likely amount, IFRS: expected value for multiple scenarios). The rule engine is configured per applicable accounting standard.

How are onerous contract provisions handled?

Onerous contract provisions per HGB Paragraph 249 Abs. 1 are maintained as a separate category. Calculating the anticipated loss requires human assessment of contract risk. The agent prepares the data basis (contract value, current costs, remaining effort).

What happens when a litigation provision proves unfounded?

The agent reviews at every close whether the prerequisites for existing provisions still hold. When the basis ceases (e.g. litigation won), the release is prepared and submitted for approval.

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.