TRM Labs Raises $70M Series C for AI Crime-Fighting Push

A blockchain analytics and artificial intelligence solutions vendor led by a former McKinsey analyst raised $70 million to expand AI-powered tools across investigations, compliance and intelligence.

See Also: Experts Offer Insights from Theoretical to the Realities of AI-enabled Cybercrime

The Blockchain Capital-led Series C round will help San Francisco-based TRM Labs pursue investigations that use orchestrated AI agents to scale blockchain forensics and compliance solutions that help analysts manage transaction monitoring, due diligence and KYC at unprecedented scale, said co-founder and CEO Esteban Castano. He said TRM Labs will continue to invest in data and threat intelligence.

“Today, defenders don’t have enough tools to keep pace with the speed and scale at which the threat landscape is evolving,” Castano told Information Security Media Group. “We’re raising this new financing to double down on our core blockchain intelligence platform and also to introduce new AI tools to cyber defenders and criminal investigators so that they can monitor and disrupt AI-scale crime.”

TRM Labs, founded in 2018, employs 369 people and has raised $220 million, having last completed a $70 million Series B extension round in November 2022 led by Thoma Bravo. The company has been led since its inception by Castano, who previously spent 17 months as a business analyst for McKinsey. TRM Labs received a $1 billion valuation in conjunction with its Series C funding (see: Cryptohack Roundup: Alleged Fraud Kingpin Deported to China).

How to Orchestrate AI Agents to Perform Investigations

Criminals have always been early adopters of technology, Castano said, first using the internet to globalize victim reach and later using cryptocurrency to move money instantly across borders. With AI, criminals can replicate and scale their operations using machines, but defenders continue to rely on tools and workflows designed for a slower, human-scale threat environment, Castano said.

“Criminals have always leveraged new technologies,” Castano said. “They leverage the internet to reach victims anywhere, then they leverage crypto to be able to move money instantly across borders. And now with AI, they have the ability to replicate themselves and scale not with headcount, but with machines.”

While TRM has used AI for years to process blockchain data, the day-to-day experience for investigators is still largely manual, Castano said, clicking through transactions and tracing flows by hand. AI-powered investigations involve orchestrating AI agents to perform the first pass of tracing and pattern discovery, allowing machines to handle scale and speed while humans focus their expertise where it matters most.

“We believe the future of investigations is not manually connecting the dots by hand, but orchestrating AI agents to build investigations for you,” Castano said. “Investigators today are point-and-clicking transactions by hand, and that approach isn’t going to scale. We traced funds through over 17,000 transactions in one case with a team working 24/7. Most companies don’t have the means to do that.”

AI agents are positioned as tools for speed, scale and task automation, he said, capable of translating natural language commands into investigative actions such as plotting transactions or pulling related data. But decisions that carry legal or ethical weight such as interpreting results, making evidentiary decisions and determining outcomes must remain with humans, Castano said.

“There’s certain domains where judgment needs to continue to be in the hands of the user, and criminal investigations is certainly one of those areas,” Castano said. “AI agents can deliver a huge scale and speed advantage to the investigator through task automation. That gives you the best of both worlds, where you have the speed and scale that AI offers, but the investigator stays in the driver’s seat.”

How Skyrocketing Transaction Volumes Upended Compliance

As stablecoins and tokenized assets drive transaction volumes into the trillions, Castano said traditional compliance models break down since financial institutions cannot hire enough analysts to manually monitor this scale of activity. TRM wants to allow a single compliance analyst to operate as if they were backed by dozens or even hundreds of analysts through the orchestration of AI agents, Castano said.

“If we’re moving to a world where trillions of transactions are happening in stablecoins and the cost of payments comes down by an order of 10 or 100, then financial institutions need a fundamentally different approach to how they monitor their transactions and assess financial crime risk,” Castano said.

As digital assets move beyond speculative crypto investing and into mainstream financial activity such as tokenized deposits and money market funds, understanding a client’s digital asset exposure becomes a core compliance obligation, he said. By providing insight into customers’ on-chain activity and exposure, TRM can help financial institutions make more informed onboarding and due diligence decisions.

“They’re not going to be able to hire their way to monitoring that risk exposure by hand,” Castano said. “Our core product question is, ‘How do we enable every compliance analyst to operate as if they were 100 compliance analysts?'”

Years ago, Castano said TRM recognized that the sheer volume and velocity of emerging threats from darknet markets to fraud schemes would overwhelm any purely human-driven intelligence operation. By applying AI to threat collection and attribution, TRM was able to map the illicit landscape at the scale of compute, delivering faster intelligence delivery, better cost efficiency and timely, actionable insights.

“We realized that the speed at which new darknet marketplaces, investment fraud schemes, child exploitation services and other threats were expanding would outpace any organization’s ability to track those threats by hand,” Castano said. “We started investing in AI capabilities internally to be able to map the threat landscape at the scale of compute rather than the scale of headcount.”