Ritholtz picks Hadrius to Police Firmwide Claude Rollout

Ritholtz Wealth Management, the New York registered investment adviser that has grown to a company-stated $9.4 billion in assets, has hired compliance startup Hadrius to monitor and archive every employee interaction with Anthropic's Claude, clearing the regulatory path to put an enterprise chatbot in front of its entire workforce. The arrangement gives Ritholtz's compliance team a logged, searchable record of AI usage that can be produced for an examiner, and it makes the RIA an early test case for how wealth managers supervise generative AI under rules that regulators have yet to write.

The deal matters less for its size than for what it signals. Two-thirds of investment advisers are already using AI in some form, yet the supervisory framework governing that use remains largely improvised. Ritholtz and Hadrius are effectively building the oversight model in public, ahead of prescriptive guidance from the Securities and Exchange Commission or the Financial Industry Regulatory Authority.

What does the Hadrius deal actually do?

Hadrius will monitor Claude Enterprise activity and data access across Ritholtz and capture user interactions with the chatbot into an audit-ready archive. The system runs on Hadrius' integration with Anthropic's Claude Compliance API, which Anthropic launched in May 2026 alongside 28 enterprise security and compliance integrations. That API exposes two categories of data: conversation content from Claude Enterprise, meaning chats, uploaded files and project material, and activity event logs covering logins, admin actions and configuration changes.

Hadrius sits on top of that feed, applying real-time usage metrics to check that employee use of the large language model conforms to data-security standards, then archiving records in line with SEC and FINRA recordkeeping obligations. The company describes the module as the first AI governance product built specifically for financial services. Anthropic's own documentation lists Hadrius among the compliance partners integrating with the Compliance API, capturing Claude conversations and activity into a books-and-records archive for SEC- and FINRA-regulated firms.

As part of the rollout, Ritholtz will also move its non-AI communications archiving onto Hadrius, consolidating email and other channels alongside AI activity in a single system rather than running AI oversight as a separate silo.

Why is AI supervision such a grey area?

Regulators have made clear that existing supervisory and recordkeeping duties extend to AI, but they have not published the specific monitoring standards firms should apply. That leaves compliance officers deploying tools for which no rulebook yet exists, and carrying the personal liability if supervision falls short. The practical problem is one of scale. When AI generates communications, marketing and trade-related content, the volume quickly outpaces what human reviewers can process, and the regulatory expectation to review, document and retain does not shrink to match.

Hadrius' pitch is to treat AI activity the way firms already treat email: logged, reviewable

and ready for examination. Thomas Stewart, the company's co-founder and chief executive, framed the tool as a way for firms to adopt AI without stepping outside their compliance obligations, arguing that oversight, not abstinence, is what lets advisers use the technology. Ritholtz chief executive Josh Brown positioned the decision as a bet that staff will use AI regardless of firm policy, and that the responsible response is to give them capable tools with visibility and controls attached rather than to leave usage unmonitored.

Who are the two firms behind the deal?

Hadrius was founded in 2023 by Som Mohapatra, Thomas Stewart and Allen Calderwood, a team that previously built and ran SEC-registered robo-adviser Quantbase. In July 2026 the company disclosed $27 million in combined seed and Series A funding led by CRV, with participation from Y Combinator, Pathlight Ventures and the founders of Altruist, Jump AI and FINNY. Hadrius says more than 500 financial institutions and investment firms run compliance programmes on its platform, and that it reduces manual compliance review by 70%.

Ritholtz has been one of the fastest-growing independent RIAs in the United States,

expanding from $2.2 billion in assets in 2022 to $7.7 billion by early 2026 without taking outside capital, according to industry reporting, on an average of roughly $1 billion of net growth a year. In February 2026 the firm rolled out an employee-led succession plan that handed ownership stakes to 29 staff. Its co-founders, Brown and Barry Ritholtz, are among the most visible commentators in US wealth management through their media and podcast output, which makes the firm's early public embrace of governed AI a notable signal to peers.

How does this fit the wider compliance-tech shift?

The rollout lands in a market where AI governance is moving from optional to expected. Anthropic's Compliance API launch drew integrations from established security names including CrowdStrike, Microsoft Purview, Okta and Zscaler, evidence that AI usage is being folded into the same governance stack firms already run for email, Slack and other SaaS channels. What differentiates the Hadrius module is its narrow focus on the recordkeeping rules specific to broker-dealers and investment advisers rather than general enterprise security.

For competitors in the compliance-technology space, from incumbents such as Smarsh and StarCompliance to newer entrants, the deal is a marker of where demand is heading. The supervisory question is no longer whether AI activity must be captured, but which vendor captures it in a form an examiner will accept.

Why This Matters to FinanceX Readers

Every regulated wealth firm now faces the same decision Ritholtz has made explicit: staff will use AI whether or not there is a policy for it, so the practical choice is between governed adoption and unmonitored risk. The firms that move first are setting the de facto standard that later SEC and FINRA guidance is likely to codify, and they are doing so on their own terms rather than retrofitting controls after an enforcement action.

For investors and operators, the signal is that AI supervision is becoming a distinct budget line and a distinct vendor category. Compliance is one of the largest and least automated cost centres in financial services, and the race to bring AI activity into audit-ready recordkeeping is opening a market that barely existed 18 months ago. Watching which oversight models regulators ultimately bless will tell you a great deal about where compliance-technology spending flows next.