news & insights
back to all news
At the SAP Concur Fusion Exchange Paris 2026, Arago took to the stage at a conference that confirmed a key trend: expense management is undergoing a fundamental transformation with the rise of AI. Alongside Equans, an Arago client for its expense and travel expense management, Julien Martinet, a T&E consultant at Arago, spoke to share practical insights on the challenges of governance, architecture and adoption surrounding SAP Concur.
A look back at this presentation, which highlighted a reality that can no longer be ignored: compliance costs are no longer merely an accounting issue, but a genuine business risk.
Expense report fraud is not a new issue. Personal meals claimed as business expenses, receipts submitted twice, whether by oversight or opportunism, inflated amounts: these practices cost companies millions each year and expose employers to financial penalties and audits. In the vast majority of cases, this is not organized fraud, but rather a gradual, almost routine drift, made easier by the lack of systematic controls.
One figure shared during the conference was particularly striking: some large organizations check only 4% of their expense reports, often on a random basis. The outcome is predictable, very few detections and many blind spots.
What is changing today is the ability of generative artificial intelligence to produce visually convincing documents. During the conference, a simple demonstration made the point clear: in less than thirty seconds, ChatGPT generated a credible restaurant receipt, complete with logo, address, date, and consistent amounts. The AI even suggested ways to make the fake look even more realistic.
This is no longer a theoretical threat, but an operational reality that Finance teams and management controllers must be prepared to address today.
When an employee submits a personal meal as a business expense, the company reimburses a cost that is not related to professional activity. From a tax perspective, this is considered disguised remuneration: amounts that should have been subject to social security contributions but were not. In the event of an audit, the company is on the front line, even if the fraud was committed by an individual employee.
Tax non-compliance, such as improperly reclaimed VAT or non-eligible expenses, adds an additional layer of risk in multi-country environments, where rules vary from one jurisdiction to another.
In sectors exposed to public procurement, the consequences can be far more severe. Equans, a subsidiary of the Bouygues Group specializing in services for local authorities and an Arago client, stated this very clearly during the Concur Fusion Exchange: proven non-compliance can lead to a ban on bidding for public contracts. For a business unit where certain contracts represent several billion euros in revenue, the issue goes far beyond accounting.
Expense fraud, if made public or uncovered during an external audit, can also seriously damage a company’s brand image and credibility.
Checking a sample of expense claims is no longer sufficient to identify large-scale or repeated fraud. When faced with AI-generated supporting documents, technology becomes the only credible solution.
Several complementary approaches are emerging:
Tools such as Detect by Oversight, which are integrated into the SAP Concur ecosystem, now enable:
The challenge is no longer to carry out more audits, but to audit more effectively: to identify genuine anomalies, prioritise risks and free teams from checks that add no value.
Expense compliance is not built at the moment of approval alone. It is established upstream, maintained continuously, and reinforced downstream. During the Concur Fusion Exchange, Julien Martinet, T&E Consultant at Arago, presented a four-level model to structure a robust and balanced compliance approach:
Compliance starts with understanding. An employee who is unfamiliar with their company’s expense policy cannot be expected to comply with it. Conversational AI tools (such as Joule, SAP’s assistant) now make it possible to answer users’ questions about applicable rules in real time, without mobilizing Finance teams. SAP Concur administrators can also create control rules directly using natural language—a major step forward in simplifying policy management.
When an employee enters an expense report, intelligent rules can block or trigger alerts in real time: amounts exceeding authorized thresholds, non-eligible expense categories, or missing mandatory receipts. When properly configured, this level of control significantly reduces errors and unintentional policy violations.
Before an expense report is approved and reimbursed, targeted controls can be applied to critical items. This is where tools designed to detect suspicious receipts or obvious anomalies come into play. The objective is clear: stop what must be stopped—without unnecessarily slowing down the approval workflow.
This is where the true power of data comes into play. Based on a consolidated dataset of all transactions, behavioral analysis reveals trends, recurring patterns, and risk-prone behaviors that are invisible at the individual expense level. Reporting and financial intelligence tools enable real-time risk monitoring and decision-making.
The challenge lies in finding the right balance: not blocking everything upfront at the expense of user experience, but not letting everything pass through with nothing more than random audits. Effective compliance means applying the right control, at the right time, to the right elements.
Companies that have taken the plunge into an automated compliance system are seeing a number of tangible benefits:
This is precisely what Equans, a subsidiary of Bouygues (90,000 employees, €19 billion turnover, 2,200 locations worldwide), has implemented with the trio of solutions SAP Concur + Detect + Extract It, integrated by Arago.
Choosing Arago is not just about implementing a tool. It is about building a sustainable and robust expense compliance framework.
Generative AI has exposed a structural weakness in traditional control mechanisms. Expense report compliance can no longer rely on trust or sampling: it must be underpinned by technology, governance and data.
At Arago, we firmly believe that well-designed expense compliance can drive operational performance and provide peace of mind for finance departments.
Is your system up to the task of managing new risks? The Arago teams can assist you with auditing, configuring and optimising your SAP Concur environment.
Yes. Tools such as ChatGPT can generate a visually convincing restaurant receipt in under 30 seconds (complete with a consistent logo, address, date and amount). Traditional visual checks are no longer sufficient: only by analysing the metadata and file structure can one detect that the document has been generated artificially.
Every year, it results in millions of euros in losses, often in subtle ways: personal meals being claimed, duplicate claims, and inflated amounts. The risk is exacerbated by very limited checks: some large companies only verify 4% of their expense claims, on a random basis.
These include financial, tax and criminal liabilities. The reimbursement of non-business-related expenses exposes the company to social security contribution adjustments and VAT that has been incorrectly reclaimed. In sectors subject to public procurement, proven non-compliance may result in a ban on tendering for contracts.
SAP Concur offers several levels of control: intelligent pre-submission rules, pre-payment verification and post-payment auditing using behavioural analysis. The solution enables 100% of transactions to be analysed and nearly 80% of checks to be automated.
In four stages: training staff on expense policies, integrated pre-submission checks in SAP Concur, pre-payment verification of high-risk items, followed by a data-driven post-payment audit. The aim: to protect the company without compromising the employee experience.
Yes. As a certified SAP Concur partner, Arago supports its clients with expense compliance audits, the configuration of controls, the integration of SAP Concur, and the implementation of Extract It for data traceability, including across multiple countries.