TRM Labs has launched Co-Case Agent, a new AI investigative assistant embedded directly inside TRM Forensics, the company’s flagship crypto investigations platform. The product is available immediately to TRM Forensics customers across crypto businesses, law enforcement, financial institutions, regulators and national security agencies, and TRM says it comes at no additional cost.
The pitch is not that AI will replace investigators. It is that crypto investigations are now moving faster than human teams can reasonably keep up with on their own, especially as cases span multiple blockchains, legal regimes and criminal typologies. TRM says Co-Case Agent is designed to shorten that gap by turning natural-language prompts into investigative actions such as tracing fund flows, auditing case graphs and suggesting next steps while keeping the investigator in control.
An AI copilot built directly into the case workflow
The most important detail in the launch is where the product sits. Co-Case Agent is not described as a separate chatbot or a generic productivity tool. It is embedded inside TRM Forensics and is meant to work directly on active crypto investigations. According to TRM, users can ask it in plain language to trace funds, run graph audits, surface the next likely move in a case and document investigative work in a way that can later support suspicious activity reports and legal proceedings.
That matters because the product is being framed as part of the actual investigative workflow, not as a sidecar assistant. TRM’s argument is that investigators are losing time not only to the complexity of blockchain tracing but also to the manual effort required to move from one investigative step to the next before illicit funds are off-ramped or hidden.
Built for defensibility, not black-box output
TRM is making explainability one of the core selling points. The company says Co-Case Agent follows a “glass-box” model, meaning it suggests, explains and documents its actions rather than silently changing graphs or producing unsupported conclusions. Every recommendation, TRM says, can be traced back to the underlying graph and the company’s attribution data.
Just as important, TRM says every interaction with the agent is written to an immutable audit log. That is a notable feature because the end product of many investigations is not just internal intelligence. It is something that has to stand up in a SAR filing, a court exhibit or a regulatory response. TRM is clearly trying to position Co-Case Agent as an AI tool that can operate inside evidentiary and compliance-heavy environments rather than only inside exploratory research.
Human judgment is still the final layer
TRM repeatedly stresses that investigators remain in control. The company says Co-Case Agent executes actions at the investigator’s direction and is designed to accelerate expert judgment, not replace it. That distinction is central to the launch because many compliance and law-enforcement users are likely to be more interested in assisted decision-making than in fully autonomous case conclusions.
Why TRM thinks the timing is right
TRM is tying the launch to a broader shift in crypto-enabled crime. In the announcement, the company says illicit crypto volume reached $158 billion in 2025, AI-enabled scams surged 500% year over year, and criminal networks are increasingly able to move and obscure funds across chains in seconds. Those figures are presented by TRM as part of the case for AI-driven investigations in an AI-shaped threat environment.
The broader message is that investigative teams are facing a throughput problem. TRM says one investigator should be able to work with the throughput of an entire team, or at least operate far more quickly than traditional manual workflows allow. Whether that claim holds up in day-to-day use will depend on adoption and accuracy, but it explains why TRM is launching an agent now instead of treating AI as a future add-on.
What is live now — and what is still ahead
TRM says Co-Case Agent is available now to all TRM Forensics customers beginning March 25, 2026, with additional capabilities to be released on a rolling basis. That means this is a real product launch, not just an early concept announcement. At the same time, the release does not describe the upcoming features in detail, does not publish benchmark accuracy figures, and does not say how much case time the product actually saves in production environments.
There is also no detailed breakdown yet of which investigative actions are fully automated, which are semi-assisted, or how broadly the agent works across all blockchains and case types supported by TRM. So the product scope is clear at a high level, but several practical performance questions still remain open.
Why it matters for crypto
- It shows blockchain intelligence firms are moving AI from research and narrative into the actual case workflow.
- It suggests the next stage of crypto investigations will be defined as much by speed, explainability and auditability as by raw tracing capability.
- It reinforces that AI is now part of both sides of the crypto-crime equation: criminals are using it, and investigators are starting to use it too.
- It may raise expectations across the compliance and investigations market for agent-style tools that can support SARs, court evidence and regulatory responses without becoming black boxes.
What to watch next
- Whether TRM publishes real-world productivity data showing how much faster Co-Case Agent makes investigations.
- Which additional capabilities arrive in the rolling releases TRM has promised.
- Whether rival blockchain intelligence and compliance firms respond with similar embedded AI case agents.
- Whether customers trust agent-assisted outputs enough to rely on them in live enforcement, compliance and prosecutorial workflows.