AI Automation

AI in Finance: Controlling, Liquidity Planning, and Risk Management for Mid-Sized Companies

How AI transforms controlling, liquidity planning, and risk management in mid-sized companies—practical examples, tools, and a concrete getting-started guide.

Author
SIMO GmbH
Published on
Reading time
12 minutes

Finance departments in 2026 face triple pressure: rising market volatility, tightened regulation, and an acute skilled labor shortage that is thinning out traditional controlling teams. At the same time, the KfW study from February 2026 shows that 20 percent of all mid-sized companies in Germany now use AI—five times as many as in 2018. Yet according to Deloitte Finance Trends 2026, only 21 percent of companies with AI implementation achieve a clearly measurable value contribution. The gap between technology adoption and real benefit is therefore real. This article shows where AI concretely applies in controlling, liquidity planning, and risk management, what results mid-sized companies achieve with it, and how to get started.

Why Finance Departments Have the Greatest AI Leverage

The finance function is the area with the highest automation potential in the entire company. There are three structural reasons for this.

First: Data richness. Finance departments work with the most structured and complete data in the company. Every transaction, every payment flow, every posting generates a digital record. That is exactly the prerequisite AI systems need to identify precise patterns and create forecasts. With a digitalization rate of 82 percent, the finance area already has a lead over other business departments according to Haufe.

Second: Repeatability. Controlling processes like monthly closings, actual-versus-budget comparisons, and liquidity reports follow recurring patterns. Exactly these repetitive, rule-based workflows are predestined for AI-powered automation. The Horvath CFO Study 2025 shows: 54 percent of companies perceive an enormous increase in workload in the finance function but can hardly add staff.

Third: Business-critical decisions. Liquidity shortfalls, currency risks, cost explosions—financial decisions have immediate implications for a company’s viability. AI can not only analyze faster here but also identify risks that are simply overlooked in manual processes.

The Deloitte Finance Trends 2026 study, based on a survey of 1,326 CFOs worldwide, confirms this finding: 63 percent of finance leaders report comprehensive AI implementation. But the measurable benefit lags behind. This is not due to the technology itself but to the lack of strategic application.

Controlling: From Rearview-Mirror Analysis to Forward-Looking Management

AI-Powered Controlling in Practice

Traditional controlling works backward: monthly reports document what happened. AI-powered controlling flips the perspective. Instead of preparing past figures, it forecasts future developments and delivers actionable recommendations in real time.

This works through three mechanisms:

  • Automated variance analysis: AI systems continuously compare actual data with planned values and identify deviations the moment they occur—not in the next monthly report. Gradient-boosting models and deep-learning-based time series architectures also detect nonlinear relationships that classical regression models miss.
  • Predictive forecasting: Instead of linear projections, AI analyzes historical data, seasonal patterns, and external factors simultaneously. KPMG reports in its Corporate Treasury News from February 2026 that probabilistic forecasting methods enable the representation of ranges and scenarios rather than single point values. This significantly improves risk management.
  • Intelligent reporting: AI not only generates reports but interprets them. Anomalies are automatically flagged, trends commented on, and action options suggested. The controller evolves from data collector to strategic advisor.

What This Means for the Mid-Market

Mid-sized companies benefit disproportionately because they typically have leaner controlling teams than corporations. When a two-person controlling team spends 30 percent of its time manually creating standard reports, AI automation targets exactly where the bottleneck is greatest.

The KfW study confirms: among larger mid-sized companies with more than 50 employees, 36 percent already use AI. But even among small companies with fewer than five employees, the rate stands at 19 percent. The entry barrier is continuously declining.

Liquidity Planning: AI Makes Cash Flow Predictable

Why Traditional Liquidity Planning Reaches Its Limits

Liquidity planning is one of the most demanding tasks in finance. It must simultaneously account for incoming payments, liabilities, seasonal fluctuations, currency effects, and unforeseen events. In practice, many mid-sized companies work with Excel spreadsheets based on experience values and manual estimates.

KPMG describes in its article series on liquidity planning (December 2025 to February 2026) the typical challenges: fragmented data landscapes, predominantly manual processes, limited transparency, and the pronounced high volatility of central influencing factors. Even small deviations in posting logic or currency conventions can mean that data is not directly comparable.

How AI Transforms Liquidity Planning

AI-powered liquidity planning uses predictive analytics—forecasting models based on historical payment data. Unlike generative AI such as ChatGPT, these are specialized machine learning algorithms that identify patterns in payment flows and calculate forecasts from them.

The concrete advantages according to current studies:

  • Metric · Without AI · With AI · Improvement
  • Forecast accuracy · 65-75 percent · 85-95 percent · Up to 30 percent
  • Time required for planning · 2-3 days/month · 2-4 hours/month · Approx. 80 percent
  • Planning horizon · 4-8 weeks · 12-26 weeks · 3x longer
  • Anomaly detection · Manual, reactive · Automatic, proactive · Real time instead of monthly report

Nomentia, one of the leading providers of AI-powered cash flow forecasting, describes the mechanism as follows: various algorithms identify patterns in historical payment data and use them to calculate forecasts. Once set up, AI-powered methods optimize liquidity management through more accurate forecasts, deeper insights into cash flow drivers, and less manual effort.

Practical Example: BASF Deploys AI in Treasury

At the KPMG Digital Treasury Summit 2025, Dr. Joanna Scheinker from BASF presented how the corporation uses AI in liquidity planning. The in-house AI solution creates forecasts per cash position based on historical data from the last seven to ten years and identifies recurring patterns. A key insight: deploying AI does not primarily lead to personnel savings but significantly increases the accuracy and speed of liquidity planning. At the same time, constant monitoring and regular adjustment needs were evident.

For mid-sized companies, this means: you do not need to be BASF to benefit from AI-powered liquidity planning. Providers like Nomentia, Kyriba, and Agicap offer scalable solutions that work for smaller treasury teams as well.

Risk Management: Early Warning Systems Instead of Firefighting

The New Dimension of Risk Detection

In March 2026, BizTech Magazine published an analysis that captures the paradigm shift in risk management precisely: risk management is no longer a brake but an accelerator. When risk and compliance teams know that automated guardrails function reliably, they stop acting as blockers. Deployments accelerate because trust exists by design.

For the mid-market, this is particularly relevant because dedicated risk management departments are often absent. AI closes this gap with three core functions:

Automated risk detection: AI systems continuously scan financial data, market data, and external information for risk indicators. A sudden deterioration in a major customer’s payment behavior, an unusual spike in material costs, a currency fluctuation that increases procurement costs—AI detects all of this in real time, while conventional systems only flag it in the next monthly report.

Scenario analyses and stress tests: Instead of having only one planned value, companies can use AI to simulate various scenarios. What happens if the largest customer pays 30 days later? How does a 15 percent raw material price increase affect the margin? AI calculates these scenarios in seconds and delivers probabilistic results rather than single-point values.

Compliance monitoring: With the EU AI Act, whose rules for high-risk AI systems take full effect from August 2026, regulatory pressure is also growing. AI-powered compliance systems automatically monitor whether financial processes comply with applicable regulations and raise alerts upon deviations.

What Is Different in 2026: Agentic AI in Risk Management

A trend shaping risk management in 2026 is the transition from generative to agentic AI. According to BizTech Magazine, agentic systems do not just make recommendations—they act independently. They move money, approve transactions, route workflows, and interact with third parties at machine speed.

KPMG describes the current state as a transitional phase—from rule-based automation through machine learning to agentic systems. The latter promise a nearly autonomous treasury in the long term, in which AI agents independently simulate scenarios and propose measures.

Human control remains decisive. As BizTech emphasizes: a hallucination in a chatbot is embarrassing. A hallucination in an autonomous agent is an operational event—it can create fraud risks, regulatory violations, or real financial losses.

Practical Guide: AI-Powered Financial Controlling in 90 Days

The Deloitte study is clear: AI projects should begin with a specific, high-impact use case where ROI is quickly measurable. Here is a structured roadmap for mid-sized companies.

Phase 1: Analysis and Quick Wins (Day 1-30)

  • Check data quality. AI is only as good as the data it works with. Check: is your financial data consistent, complete, and in a uniform format? KPMG emphasizes that even small deviations in posting logic can distort results.
  • Identify one use case. Do not automate everything at once. Start with the area that has the greatest pain point. Typical quick wins: automated variance analyses, cash flow forecasts, or automated dunning.
  • Start tool evaluation. Define requirements: ERP connectivity, DATEV integration, GDPR compliance, scalability.

Phase 2: Pilot Project (Day 31-60)

  • Set up the pilot. Implement the AI solution in a delimited area. Example: AI-powered liquidity forecast for the next 12 weeks, based on historical payment data from the past three years.
  • Measure results. Compare the AI forecast with previous manual planning. How large is the deviation? How much time does the team save?
  • Involve the team. Train controllers and the finance team in using the new system. Explainable AI is crucial here—the team must understand why the AI makes certain forecasts.

Phase 3: Scaling and Optimization (Day 61-90)

  • Document successes. Create a business case with concrete numbers: saved hours, improved forecast accuracy, reduced error rates.
  • Scale. Transfer the insights gained to additional use cases: from cash flow forecasting to risk management, from controlling to reporting.
  • Establish governance. Define clear rules: who reviews AI results? Which decisions does the AI make autonomously, which require human approval?

Practical Example: Mid-Sized Mechanical Engineering Company Saves 35 Percent of Controlling Effort

A mechanical engineering company with 120 employees in North Rhine-Westphalia introduced an AI-powered controlling solution in Q4 2025. The starting situation: the three-person controlling team spent approximately 120 hours monthly on manual tasks—data preparation, actual-versus-budget comparisons, monthly report creation, and liquidity planning.

Results after 90 days:

  • Metric · Before · After · Savings
  • Time for monthly closing · 5 working days · 2 working days · 60 percent
  • Manual report creation · 40 hours/month · 12 hours/month · 70 percent
  • Cash flow forecast accuracy · 68 percent · 89 percent · Plus 21 percentage points
  • Detection of plan deviations · Month-end · Real time · Immediate response
  • Total controlling effort · 120 hours/month · 78 hours/month · 35 percent

The 42 hours saved per month are now used by the team for strategic analyses, investment evaluations, and preparing decision papers for management. The ROI of the implementation was achieved after eight weeks.

Frequently Asked Questions

Do we need an ERP system before we can use AI in controlling?

Not necessarily, but it helps considerably. AI needs structured data, and an ERP system provides exactly that. Those still working with Excel can still get started—but the effort for data preparation is higher. The KPMG analysis shows that heterogeneous system landscapes represent the biggest hurdle for AI in treasury. A pragmatic approach: start with the data you have and improve data quality step by step.

How high are the costs for AI-powered controlling in mid-sized companies?

The range is broad. Cloud-based solutions for liquidity planning start at 200 to 500 euros monthly. More comprehensive controlling platforms with predictive analytics run at 1,000 to 3,000 euros monthly. Enterprise solutions are individually priced. What matters is the ROI: with time savings of 42 hours per month and an internal hourly rate of 55 euros, the monthly savings amount to 2,310 euros—even a solution costing 1,500 euros per month pays for itself within three months.

Can AI in finance be used in compliance with GDPR?

Yes, if the right framework conditions are in place. Watch for: server location in Germany or the EU, encryption during transmission and storage, clear rules on data usage for AI training, and auditability of all AI decisions. The EU AI Act, whose rules for high-risk systems take effect from August 2026, additionally requires transparency and human oversight for AI systems in finance.

Does AI replace the controller?

No. The Deloitte study shows that nearly 70 percent of CFOs view AI as a tool that complements human capabilities. The controller’s role shifts: away from data preparation, toward strategic interpretation and decision support. KPMG describes this as “human-in-the-loop”—humans remain in the decision process but take on a more strategic function.

At what company size does AI in finance make economic sense?

The KfW study shows that 19 percent of companies with fewer than five employees already use AI. For finance specifically: from a monthly transaction volume of 200 to 300 postings and a controlling effort of more than 20 hours per month, deployment becomes economically sensible. What matters is not company size but data quality and the willingness to change processes.

References

Tags

  • SMEs
  • Automation
  • ROI
  • Mid-Market
  • Process Automation

Back to the overview

Business Data Strategy for your company

From the target state to Delivery Supervision. We advise you and enable your organization.

Book a discovery callView our advisory work