On September 14, Anthropic launched Claude for Financial Advisors with live connectors into several platforms your advisors likely already touch. That same day, the SEC published a risk alert on how firms fall short in their annual compliance reviews. Taken together, these two events are the clearest signal yet that AI is moving from something advisors experiment with to something embedded in the tools advisors use every day, including the solutions that touch client communications and investment advice. 

For InvestorCOM clients, the question isn’t whether your advisors will use AI. Increasingly, they already do, or will soon. The question is what your firm needs to have in place around it, particularly anywhere AI-generated content is heading toward an investor. 

Every piece of AI-assisted communication needs a human in the loop 

This is the single most important control, and it’s also the one that erodes quietest. AI-generated summaries, meeting notes, and portfolio commentary tend to read fluently enough that they get skimmed instead of checked, and a wrong figure or an unsupported claim slips through into something a client reads. 

The standard doesn’t change because a machine helped draft it. Every factual statement needs to trace back to a source your firm can produce. Every figure needs to be verified, not assumed correct because it looked right. Nothing should imply a guaranteed result. Required disclosures still need to be there. And that review needs to be documented a review nobody can evidence is, to a regulator, a review that didn’t happen. 

Important to remember, Rule 206(4)-1 obligations sit with the firm, not you solution provider. For example, if an AI solution screens a communication, misses something, and it goes out to a client, the firm answers for it, not the solution provider.  

AI-assisted advice and communications are still records 

Doubly important, if an AI solution drafts a client email, a meeting summary, or an analysis that informs advice, that’s a record under Rule 204-2 and Exchange Act Rule 17a-4. Meeting notes generated by an AI assistant can feel like scratch paper, but if they document a client interaction or inform what an advisor recommends, they need to sit inside your retention program like anything else your firm produces. 

Know your data flow 

Every AI connector into a custodian, CRM, or portfolio system is effectively a new service provider with access to client data. Before any of these solutions are switched on for advisors, firms should be able to answer:  

1. What happens to client data?
2. Is it used to train models?
3. Where is it processed and stored
4. What can the connector actually see? and,
5. What happens to that data if the firm terminates the agreement? 

What this means in practice 

None of this is a reason to avoid AI-assisted solutions as they can mean better-prepared advisors and faster turnaround on routine work. It is however worth asking your advisor or your firm’s compliance contact a few direct questions:  

1. Is there a human reviewing anything AI-generated before it reaches a client?  
2. Does the firm have a written policy covering AI use in client communications and advice?  

The absence of clear answers is itself informative. 

Stay in the loop 

AI is moving quickly into the advice workflow, and InvestorCOM is building the guardrails and tools to match. In the coming weeks, we’ll be sharing more about how our own AI-enabled solutions are designed with human review, compliance, and recordkeeping built in from the start. Watch this space for those announcements.