Marble wants AI agents to handle compliance's grunt work. The harder job is getting banks to trust them.

Ask someone who works in bank compliance to describe the job, and you'll probably hear about the queue. Alerts arrive by the thousands, flagged by monitoring software tuned to catch anything that might be money laundering or fraud. Nearly all of them turn out to be nothing. Somebody still has to check.

Arnaud Schwartz, CEO and co-founder of Paris-based Marble, describes analysts "checking thousands and thousands of follow-ups on events, knowing that 95% of them are false positives." That grind has become one of the highest costs in banking. According to Marble, fraud and anti-money laundering work, known in the trade as FRAML, can absorb up to 10 to 15% of headcount at banks and fintechs.

Schwartz traces the problem back 15 years. "Before 2010, compliance was basically open bar," he said in an interview with The French Tech Journal. What followed was a long run of new regulations and new pressure, and the industry answered by staffing up: "everyone hired tons of people, tons of people to manage that." He thinks that approach has hit its limit. "We reached a point where it's no longer human manageable anymore."

Marble's pitch is that software, and increasingly AI agents, should take the repetitive part of that work off analysts' desks. Investors are buying in. The startup announced today that it has raised a €6.5 million Series A led by Smartfin, with participation from ADNEXUS and follow-on money from existing investors Passion Capital, 42Capital, and Hexa. That brings Marble's total funding to €9 million.