AI in audit report review: dhpg and Cortea agree on technology partnership
dhpg, one of Germany's largest mid-market audit firms, and Cortea agree a long-term technology partnership: AI agents for the audit, built on dhpg's own methodology and live at every office.
Bonn/Berlin, 15 September 2026: dhpg, one of the largest mid-market audit and advisory firms in Germany, and Cortea have agreed a technology partnership for the use of AI agents in the financial statement audit. Since the joint development and testing phase began in February 2026, more than 450 audit reports have been reviewed through Cortea. Since August, AI-supported audit report review has been part of dhpg's standard process.
The demands on audit quality and documentation rise year on year. Quality assurance can no longer be achieved through additional staff and further rounds of checks alone. With AI agents, a technology is available for the first time that can apply professional requirements consistently to every single report without compromising the traceability of the audit procedures.
Seizing the opportunities of AI without changing the firm's own methodology
This is precisely where the two partners meet, coming from different directions. dhpg is using the opportunities AI offers for audit quality without adapting its own methodology to an off-the-shelf product. Cortea develops agents that are built on the standards of audit firms rather than on a general-purpose model. That is why the result was a partnership rather than a licence agreement. The collaboration is designed for the long term.
dhpg's technical departments define the review questions, checklists and model audit reports the agents work from, and they decide how the agents are developed further. The joint roadmap includes ten further agents for other phases of the financial statement audit, with the aim of an audit supported by AI from end to end.
"For mid-sized businesses, what counts is personal service tailored to their very specific needs, even as technology changes the process of the annual audit. That is why, with audit quality and the opportunities arising from the use of AI in mind, we opted for a development partnership rather than a finished product: the professional requirements come from us, and Cortea develops the agents around them," says Andreas Blum, Managing Partner at dhpg.
AI agent covers both formal and substantive aspects
"Cortea is now part of our standard process before a report goes to the reviewer, and that applies at every one of our offices. The review questions behind it are our own. It allows us to raise the quality of our reports early in the process, which frees up time for the more demanding parts of the review," says Hagen Müller, German Public Auditor, Tax Advisor and Senior Partner at dhpg.
The audit report review covers audits of annual and consolidated financial statements as well as compilation reports, each based on dhpg's model audit reports. The agent covers both formal and substantive aspects.
The audit report review agent flags every finding with a rationale and a reference to where it appears in the report. The assessment is made by the auditor. Human report review remains a mandatory part of the process. The agent supports the auditor in exercising their professional judgement; it does not replace it, and it does not alter any audit procedures required under the auditing standards.
"From the outset, our vision has been to build AI agents for quality, all the way to a financial statement audit supported by AI from end to end. For us, partnerships are the principle, not the exception: we work with the most innovative audit firms internationally. In Germany, dhpg has been one of them for us from the very first week: the platform comes from us, the methodology and the processes from dhpg. The result is an agent that applies dhpg's own quality standard," says Valentin Neumann, co-founder and CEO of Cortea.
For the audit profession, the significance lies in breadth and duration: not a pilot in a single audit area, but a partnership in which dhpg's audit methodology is transferred into AI agents step by step.