For several years, AI in wealth management has been discussed mostly in terms of potential: better investment analysis, greater efficiency and more personalization. The launch of Claude for Financial Advisors gives us something more tangible — an early look at what an AI-enabled wealth management operation could actually look like. And for family offices, the most important part may have little to do with AI making investment decisions.

The real problem may be fragmentation. A family office rarely manages a single portfolio. One family can have personal accounts, trusts, companies, private equity holdings, real estate and assets across several custodians. Different generations may have different objectives and structures. The information exists, but it often exists in different systems.

Claude for Financial Advisors is interesting because Anthropic is attempting to put an AI layer across this fragmented infrastructure. The emerging ecosystem spans custodial data, CRM, portfolio reporting, financial planning and alternative investments. Integrations announced by Anthropic include Schwab Advisor Services, Wealthbox, Addepar, Orion, SS&C Black Diamond and iCapital.

What could this mean in practice? Before a family meeting, Claude can combine CRM history with portfolio and planning information to prepare a client briefing. Portfolio tools can help identify drift, concentrated positions or tax-loss harvesting opportunities and prepare scenarios for professional review. Through iCapital, private-market holdings, valuations, transactions and performance can become part of the same AI-assisted workflow as the rest of the portfolio.

After a meeting, Claude can turn the discussion into structured notes, identify follow-up actions and update CRM workflows. None of these activities is revolutionary on its own, the interesting part is that they can increasingly happen through the same interface.

AI is not replacing the wealth-tech stack

This distinction matters. Claude is not replacing custodians, portfolio platforms, CRMs or alternative-investment infrastructure. And it certainly is not replacing professional judgement. The underlying systems remain the sources of financial and client data. Claude sits above them, helping professionals retrieve, connect, interpret and act on that information. Anthropic itself draws a clear boundary: its financial-advisor tools can gather information, analyse it, prepare scenarios and draft materials, while the advisor remains responsible for judgement, recommendations and approval.

For family offices, this could be particularly important. The first major impact of generative AI may not be better investment decisions. It may be reducing the information fragmentation surrounding those decisions.

If this architecture works, the implications go beyond saving time on meeting preparation. A smaller family-office team could potentially manage greater complexity. Senior professionals could spend less time moving between systems and more time interpreting information, speaking with family members and making decisions.

But a new intelligence layer also creates new questions. How do we audit AI-generated analysis? Which calculations must remain deterministic? How should permissions work across trusts, entities and generations? And who is responsible when AI retrieves the right information but interprets it incorrectly?

These questions are far from resolved.

At CGL, this is exactly the direction we are exploring: how AI can connect the fragmented systems, data and workflows of family offices into a more intelligent operating model.