RIA AI Adoption Is Showing Up in the Filings

Advisor Perspectives welcomes guest contributions. The views presented here do not necessarily represent those of Advisor Perspectives.

For the past two years, much of the conversation about artificial intelligence in wealth management has been hypothetical. Will advisors use AI? Will clients accept it? Will firms allow sensitive information to flow through large language models? Regulatory filings show that those questions are starting to be answered.

AdvizorPro identified 2,720 distinct AI-related term and document combinations across 1,615 investment-adviser firm registrations. Generative AI appeared in filings associated with 1,064 firms. ChatGPT appeared across 602. Large language models appeared across 517, while 44 firms specifically used the phrase “agentic AI.”

These figures should not be interpreted as 1,615 confirmed AI adopters. Some firms describe the technology they use, while others discuss risks to their businesses, service providers, or portfolio companies. A few explicitly state that they do not use AI in investment decisions.

These disclosure footprints are nevertheless significant. AI has become material enough to appear in the documents through which advisory firms describe their services, investment methods, conflicts, and risks.

Most Firms Are Starting With Productivity

The most common examples of AI adoption are not autonomous portfolio managers. They are tools that help advisors summarize documents, prepare meetings, conduct research, and draft communications.

New England Asset Management reported that it uses Microsoft Copilot on approved devices to support operational efficiency. Its disclosure also describes restrictions against entering sensitive or confidential information into unapproved tools. The firm separately reports limited use of AI for investment research, portfolio analysis, and operational activities.

Sage Advisory Services reported that it provides Microsoft Copilot to all users for document summarization and content generation. It also uses AI-assisted meeting insights, notes, and summaries generated through Microsoft Teams.

Arax Advisory Partners described using Microsoft Copilot for research, analysis, document drafting, and communication, along with AI-powered note-taking tools for client discussions.

SK Wealth Management named ChatGPT, Otter.ai, Microsoft Copilot, and Perplexity AI among the firm-approved tools it may use for internal operations, communication, and meeting documentation.

These disclosures point to the first major stage of RIA AI adoption: reducing the administrative burden surrounding advice. Meeting preparation, documentation, research synthesis, and follow-up communication consume a substantial portion of an advisor’s day. AI can compress that work without necessarily changing who makes the recommendation.

For advisors, that creates the possibility of spending more time with clients and less time producing the materials around each interaction. For firms, it may allow existing teams to serve more households without increasing head count at the same rate.

AI Is Moving Closer to Financial Planning

Some firms are going beyond general productivity and applying AI directly to planning and research workflows.

Palmetto Coast Advisors reported that it uses Claude for Work to help draft and organize financial plans, prepare meeting agendas and summaries, and conduct general research and analysis. Its disclosure distinguishes the commercial subscription from consumer AI accounts and describes safeguards for client information.

401 Financial reported that it uses Claude by Anthropic for AI-assisted financial planning and to support the delivery of advice on specific topics.

Neptune Wealth described using Claude through a business account for wealth management, financial planning, and market and tax research. It also stated that it does not provide client data to the system.

Two Trails Financial Planning reported that it uses Claude for Work to draft narrative planning deliverables and Gemini for Google Workspace to summarize and analyze documents. Its disclosure also describes a minimum-necessary standard for information placed into AI systems.

This is a more consequential stage of adoption. AI is no longer limited to rewriting emails. It is helping assemble the research and analysis surrounding financial recommendations.

The advisor remains responsible for the advice, but the workflow producing that advice increasingly includes an AI layer.

A Smaller Group Is Moving Toward Agentic Advice

The most significant disclosures come from firms using AI agents to perform multistep investment activities.

Omen Advisory reported using its AI agents to construct portfolios. The firm describes existing AI-generated portfolios that are reviewed by its personnel, as well as portfolios created from client prompts. Omen also stated it uses AI agents to generate and execute transactions for recommended portfolios.

Silk describes a platform where AI agents built on large language models autonomously research, formulate, and execute investment strategies in U.S. equities. The firm also acknowledges that model reasoning may not be fully transparent and that identical inputs can produce different outputs at different times.

Deem Asset Management describes an internal platform that evolved from a retrieval-based research system into a skills-based, agentic AI architecture. The technology can generate analytical insights, including buy, sell, and hold recommendations, while the firm also describes supervisory review, governance, and limits around client suitability.

These firms are outliers, but their disclosures offer a preview of where the market could move. AI is progressing from assisting advisors with tasks to selecting tools, completing workflows and, in limited cases, participating in investment decisions.

That transition changes the risk profile. A drafting assistant can produce an inaccurate sentence. An investment agent can produce an inaccurate decision and act on it.

Governance Is Becoming Part of the Product

The most effective disclosures do more than name a tool. They explain the controls surrounding it.

Signify Wealth reported that its approved use of Claude Enterprise is governed by human review, a data-processing agreement, restrictions on model training, encryption, and breach-notification commitments.

Ashcroft Green Advisors reported that AI-generated content is reviewed and verified by investment professionals before it is used in client communications or investment decisions.

Liberty Wealth Management reported that AI-powered meeting transcription operates within its secure environment and remains subject to the same confidentiality, cybersecurity, and vendor-oversight requirements as its other technology systems.

Other firms are drawing narrower boundaries. Kimberlite Financial Services reported that its use of Claude, ChatGPT, Gemini, and Microsoft Copilot is restricted to administrative tasks such as drafting communications, summarizing public research, and producing internal documents.

Kahn Brothers Asset Management discusses chatbots, AI agents, and large language models but explicitly reported that it does not currently use AI in investment research or decision-making.

That distinction matters. Disclosure does not always mean adoption. Sometimes disclosure means a firm has evaluated the technology and established a boundary around it.

What This Means for the RIA Industry

The filings point toward three changes.

First, AI governance is becoming part of advisory firms’ operating models. Firms will need to know which tools employees use, what information enters them, whether vendors train on that information, and who reviews the output. A generic cybersecurity policy will not answer all of those questions.

Second, clients will expect greater transparency. They may reasonably want to know whether AI records meetings, reads financial plans, helps develop recommendations, or has authority to execute investment decisions. Firms that explain those practices clearly can turn governance into a trust signal.

Third, the advisor’s value will continue shifting toward judgment, context, and accountability. AI can summarize a document or produce a first draft quickly. It does not automatically understand a family’s history, recognize when a client is afraid to ask a question, or accept responsibility for a recommendation.

The advisors who benefit most will use AI to remove low-value work while strengthening the human parts of advice. They will also know when to challenge an output, when to disregard it, and when a workflow should never be automated.

The strongest AI strategy will be easy to explain: where the technology is used, what data it can access, who reviews its work, and who remains accountable to the client. The filings show that the RIA industry has already started working on those answers.

Source and methodology: AdvizorPro analysis of indexed investment-adviser regulatory filings as of September 4, 2026. Results were deduplicated by firm registration, matched term, and document type. Counts measure disclosure activity, not verified adoption or market share. The underlying filings, document references, and firm CRDs supporting the aggregate figures and examples are available from AdvizorPro.

Hesom Parhizkar is the co-founder & CPO at AdvizorPro.

A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.