MindBridge CFO Matthias Steinberg deeply understands the pressure on finance leaders, who are expected to leverage generative AI to optimize financial operations and drive company growth.

In his view, the core contradiction lies in the fact that the new generation of generative AI excels at creativity and language processing, while earlier machine learning technologies are adept at processing and analyzing massive data sets, creating a clear capability gap in financial scenarios.

Steinberg told CFO Dive that the hype around generative AI has raised expectations among CFOs and their finance teams, but "out-of-the-box" large language model (LLM)-based generative AI is not suitable for directly handling numbers.

"Currently, the industry is working to combine LLM systems with machine learning and other technologies to build tools that can truly handle massive data," Steinberg said. "This creates significant friction and pressure for CFOs—they need to deliver tangible results while the technology is still evolving. I think that's where the pain point lies."

MindBridge, headquartered in Ottawa, Canada, sits right at the center of this pressure with its business model, and it is not alone. When Hewlett Packard Enterprise (HPE) CFO Marie Myers decided to increase investment in its agentic AI tool "Alfred" (named after Batman's loyal butler), HPE partnered with Deloitte and NVIDIA to ensure "deterministic results"—meaning the same answer every time the same question is asked.

The non-deterministic nature of LLMs can be highly valuable in marketing, but it is concerning in finance. "We obviously want numbers to be 100% correct, and they must be deterministic—if I ask the same question to a technical system three times, I must get exactly the same answer all three times," Steinberg told CFO Dive, noting that the solution lies in integrating different technologies.

MindBridge's SaaS platform is used by auditors and other companies to detect anomalies and risks in financial data and systems. Steinberg describes it as a monitoring tool that automates the assurance of a company's books, but it does not replace enterprise resource planning (ERP) systems.

"We are not replacing ERP," he said. "We ingest data from ERP systems, and increasingly from other operational systems like booking or billing systems. We ingest the data, perform analysis, and feed results back to users in the simplest form, telling them which items may be high-risk and require follow-up."

MindBridge's users include large audit firms such as KPMG, as well as companies like Chevron. Steinberg says pricing for its analytics systems varies, with annual license fees starting in the low six figures.

Founded in 2015, MindBridge initially used unsupervised machine learning to drive analytics and is now developing a new product based on agentic AI, expected to launch within months. This new agentic "wrapper layer" will shift the user experience from a mode requiring multiple screens and analytical operations to a more streamlined interface.

"Up to now, auditors or finance professionals need to ensure data is loaded into our tool, the tool runs and generates billions of combinations, and then all results are fed back to different dashboards where users perform various slices and drills," he said. "In the future, you open your browser or tool in the morning, and you see a chat box, and you just ask: 'What do I need to do? What happened in the last 24 hours?'"

Steinberg has been the finance leader at MindBridge for nearly four years. According to his LinkedIn profile, he holds an MBA from INSEAD and a master's degree in engineering from RWTH Aachen University. Early in his career, he worked in private equity at Boston Consulting Group and Summit Partners. As CFO, he also helped German company Ionos go public before moving to Canada to join MindBridge.

For other CFOs looking to leverage AI, Steinberg advises choosing a relatively specific project to introduce AI, such as in accounts payable or investor relations, finding change agents (often younger employees), and limiting the scope of application to within the finance team.

"There's no silver bullet, and sometimes it can be overwhelming," he said. "But pick a project, and just get started."