Digital Business & Future 03.04.2026

6 min Read Time

53 percent of German companies already use AI in accounting – or are currently implementing it. That’s the finding of KPMG’s study Digitalization in Accounting 2025/2026. This figure has doubled in just two years. At the same time, the mandatory electronic invoicing (e-invoicing) requirement – effective since January 2025 – makes digitizing accounting non-negotiable. For SMEs, this presents a concrete opportunity: automation doesn’t just cut costs – it frees up capacity for strategic tasks that have long been sidelined by routine work.

The Key Takeaways

Where AI in Accounting Stands Today

KPMG’s study Digitalization in Accounting 2025/2026, based on 209 surveyed companies, paints a clear picture: AI has arrived in the finance department. 53 percent are already using AI or are in the implementation phase. 61 percent of respondents see AI as a key success factor in finance.

Development is particularly dynamic in machine learning: 28 percent of companies have firmly integrated ML into their processes. In 2023, that figure was just 15 percent. The most common use cases are automated document processing, anomaly detection in postings, and predictive cash flow analyses.

What surprises many companies: The greatest efficiency gain doesn’t come from processing speed, but from reducing queries and correction postings. If AI correctly posts 95 percent of documents, the loops between accounting, specialized departments, and suppliers – which are the biggest time consumers in practice – are eliminated. This is the real productivity gain.

For SMEs, what is NOT in the study is particularly relevant: According to the SME AI Index 2026 by Salesforce and the German SME Federation (DMB), 51.2 percent of SMEs are actively using or testing AI solutions. The gap between corporations and SMEs is closing faster than expected.

However, the study also shows sobering results: Only 11 percent of SMEs with fewer than 50 employees use AI solutions. The adoption gap exists less between industries than between company sizes. Main reasons include lack of internal expertise, uncertainty in selecting suitable tools, and fear of losing control over sensitive financial data. This is exactly where pragmatic entry comes in: don’t automate everything at once, but start with a single process that delivers immediately measurable benefits.

The Four Use Cases with the Greatest Leverage

AI in accounting isn’t a monolithic system that automates everything at once. The greatest value emerges from four clearly defined areas – each deployable independently.

1. Automated Document Processing: Incoming invoices, delivery notes, and cash register receipts are automatically recognized, categorized, and posted using OCR and machine learning (ML). Modern systems achieve over 95 percent recognition accuracy for structured invoices. The e-invoicing mandate introduced in January 2025 (ZUGFeRD and XRechnung formats) further simplifies the process, as invoice data is now machine-readable. In practice, companies report 60-80 percent time savings in document processing.

Real-world experience shows document processing is the typical entry point – because ROI is immediately measurable. A mid-sized trading company handling 500 incoming invoices per month saves roughly 40 hours monthly with 60 percent automation. At an internal hourly rate of €35, that’s €1,400 per month – or €16,800 annually. By contrast, a dedicated AI document-processing tool costs only €100-€300 per month.

2. Anomaly Detection: ML models analyze posting patterns to flag deviations: unusually high amounts, duplicate entries, incorrect account assignments, or transactions outside normal time windows. This delivers not just efficiency gains – but also strengthens compliance. Auditors increasingly view AI-supported controls favorably.

3. Cash Flow Forecasting: Predictive models analyze historical payment behavior, open items, and seasonal fluctuations to generate more accurate cash flow forecasts than traditional spreadsheets. For SMEs with seasonal business cycles or extended payment terms, this is a direct contribution to liquidity management.

4. VAT Automation: AI systems automatically verify correct tax rate application, validate VAT identification numbers, and prepare advance VAT returns. In international operations – where varying tax rates and reverse-charge rules apply – this significantly reduces error rates.

“Artificial intelligence accelerates digital transformation in accounting. What was once futuristic speculation is now everyday practice.”
– Paraphrased from KPMG’s press release on the Digitalization in Accounting study, November 2025

E-Invoicing Mandate as a Catalyst

Since 1 January 2025, all German B2B companies must be capable of receiving electronic invoices. From 2027 onward, sending e-invoices will become progressively mandatory. Formats ZUGFeRD 2.x and XRechnung deliver structured XML data that accounting software can ingest directly.

For AI-powered accounting, this is a game-changer: machine-readable invoice data eliminates the most error-prone step in today’s workflow – manual data entry. Instead of retyping documents or verifying OCR output, invoice data flows straight into the system and posts automatically.

At the same time, a new challenge arises: Companies receiving e-invoices must adapt their systems for compliant archiving. Germany’s GoBD (Grundsätze zur ordnungsmäßigen Führung und Aufbewahrung von Büchern – Principles for Proper Bookkeeping and Record Retention) require audit-proof storage. AI systems can help here too – automatically classifying, archiving, and preparing incoming invoices for tax audits.

An often-overlooked detail: The e-invoicing mandate also applies to outgoing invoices. Companies still sending PDF invoices via email must switch to structured formats by no later than 2028. Investing now in an AI-powered invoicing system solves both problems simultaneously: automated inbound processing and standards-compliant outbound issuance. The investment pays off twice.

Data Protection: The Biggest Practical Hurdle

Sixty-five percent of respondents in the KPMG study cited data protection and security as the top challenge when deploying AI in finance. Fifty-nine percent pointed to algorithmic transparency and traceability as barriers. For SMEs, this is a real concern – financial data ranks among the most sensitive corporate information.

Three questions arise concretely: First, where are the data processed? Cloud-based AI accounting solutions transmit financial data to external servers. For companies unwilling to do so, on-premise and hybrid alternatives exist. Second, who has access? The GDPR demands strict access controls and deletion concepts – even for AI training data. Third, how do I explain to my auditor what the AI does? Traceability of automated posting decisions must be guaranteed.

The pragmatic solution: AI systems in accounting should be designed as assistive tools, not autonomous decision-makers. Humans retain final approval authority for postings above a defined threshold. The AI makes suggestions; the accountant confirms them. This resolves both data protection concerns and the AI Act issue: An assistive system typically falls outside the scope of high-risk AI.

Another factor gaining relevance under the new DGG (Data Governance Act): Data quality determines AI output quality. Companies with poorly maintained master data won’t get good results from AI accounting. Data cleansing before AI rollout isn’t optional – it’s mandatory. This applies especially to chart of accounts, cost centers, and supplier master data.

Five Steps Toward AI-Powered Accounting

Getting started need not be disruptive. Here are five steps that work for SMEs with limited IT budgets.

1. Inventory Assessment: Which accounting processes are manual and repetitive today? Document capture, posting, and dunning are classic starting points with high automation potential.

2. Ensure E-Invoicing Readiness: Can your existing ERP system process ZUGFeRD and XRechnung? If not, now is the right time to upgrade – given the upcoming 2027 sending mandate.

3. Launch a Pilot Project: Select one process – for example, incoming invoices – and automate it with an AI tool. Providers like DATEV, SAP, Candis, or GetMyInvoices offer SME-friendly solutions.

4. Involve Employees: Accounting is a domain where AI anxiety runs particularly high. Transparent communication about the goal – relief, not replacement – plus robust change management, is critical.

5. Scale Up: After a successful pilot, expand to other processes: dunning, cash flow forecasting, VAT reporting. Lessons learned accelerate subsequent rollouts.

Keep realistic timelines in mind: A document-processing pilot can go live in four to six weeks. Extending to cash flow forecasting and anomaly detection typically takes six to twelve months. Fully transforming the finance department is a two- to three-year initiative. Start now, and by 2028 you’ll have a fully automated accounting function aligned with e-invoicing mandates and rising compliance demands.

Real-World Example: How a Trading Company Saves 300 Hours Annually

A mid-sized wholesale distributor of sanitary supplies – with 120 employees – processes around 800 incoming invoices monthly. Before adopting AI, document capture took an average of seven minutes per invoice: scanning, manual data entry, account assignment, and approval routing. That totaled 93 hours per month – spread across three clerks.

After introducing an AI-powered document management system, processing time dropped to two minutes per invoice. The AI identifies the supplier, extracts invoice data, suggests the correct posting, and routes approvals automatically to the responsible department head. Clerks intervene only when deviations occur from learned patterns. Result: 66 fewer hours per month – or 800 hours annually. The three clerks now dedicate freed-up time to supplier negotiations, early-payment discount optimization, and proactive cash flow management.

Investment breakdown: €250/month for software, €8,000 one-time integration with the existing ERP system, and three days of internal project management. Break-even was achieved after four months. The wholesaler is now planning its next phase: AI-driven dunning optimization – triggering payment reminders not by rigid time schedules, but based on each customer’s individual payment behavior.

This case demonstrates: AI in accounting needn’t be a multi-million-euro project. Starting with document processing is low-cost, fast to implement, and delivers immediate, measurable ROI. Prerequisites are clean master data – and the willingness not just to digitize, but truly automate, the process.

Conclusion

AI in accounting is no longer a vision of the future – it’s operational reality. More than half of German companies already use AI in their accounting departments. The e-invoicing mandate accelerates adoption by providing the foundational data infrastructure for automation.

For SMEs, getting started is simpler than many assume: A document-processing pilot can go live within weeks – and immediately reduce manual workload. Those who persist and gradually automate additional processes transform accounting from a cost center into a strategic control unit. The convergence of e-invoicing mandates, falling software costs, and growing skills shortages makes 2026 the ideal launch window. The technology is mature, regulatory frameworks are in place, and competitors are already moving. Companies that delay automation risk competing against those who’ve already made the leap.

Frequently Asked Questions

Does AI replace the accountant?

No. AI automates repetitive tasks like document capture and posting – but it doesn’t replace professional judgment for complex scenarios. The accountant’s role shifts from data entry to data analysis and strategic advisory.

How much does AI-powered accounting cost for an SME?

Cloud-based solutions start at €50-€200/month for small businesses. Enterprise-grade solutions with ERP integration cost significantly more. ROI comes from time savings: 60-80 percent reduction in manual document processing effort.

Is AI-powered accounting GDPR-compliant?

Yes – if configured correctly. Critical factors include server location (EU), granular access controls, deletion policies, and a valid data processing agreement. On-premise solutions offer maximum data control.

What is the e-invoicing mandate?

Since January 2025, all German B2B companies must be able to receive electronic invoices in ZUGFeRD or XRechnung format. From 2027, sending e-invoices becomes progressively mandatory. PDF invoices without structured data no longer qualify as e-invoices.

Which vendors suit SMEs?

DATEV (for tax-advisory-aligned solutions), SAP Business One (ERP-integrated), Candis (specialized in invoice processing), GetMyInvoices (document management), and Lexware (entry-level solution). Vendor choice depends on your existing ERP system and desired scope of automation.

Will auditors accept AI-generated postings?

Yes – if traceability is ensured. AI systems must log the rationale behind each posting suggestion. Auditors increasingly welcome AI-supported controls, as they operate more consistently than manual sampling.

Header Image Source: Nataliya Vaitkevich / Pexels