Claude's advisor plug-in closes the integration gap and opens a meter

Anthropic has announced an advisor-specific plug-in for Claude that draws on data from many of the key tools in an advice firm's stack, the lead item in Kitces' weekend reading for planners. The pitch is the one advisors have been chasing since their first stack audit: planning, CRM, and reporting data in one place, without a separate login and a manual export between each pair.

Kitces frames the plug-in as a potential fix for integration across a firm's tool set, then flags what it leaves open—what advisors do with the hours it hands back, and whether those hours earn a positive return against the usage-based fees Anthropic charges for the tool.

Connectivity was never the hard part: the connectors are the proof of concept, and the practices that never wrote down a baseline will have no way to know whether the tool moved anything. A plug-in that reaches across many systems makes the integration question recede and leaves the measurement question exactly where it stood.

The second of Kitces' questions is the one that decides whether the first matters, because the fee is charged either way. Schwab's exclusive Claude integration—which this publication reported at $240 a year per seat across 16,000 RIAs on its custody platform—put a price on AI seats before most firms had a view on what an AI seat produces. Usage-based AI clauses are already turning up in wealthtech contracts, passing compute costs through to advisory firms whether or not anyone switches the feature on. Put a connector in front of most of the stack and the meter stops being one line on one renewal and becomes a line on several.

Buy it if the firm can already put a number on the hours its integration work eats each month; wait if it cannot, because that cuts against how most practices shop for software, which is by demo. The demo flatters by construction—data moves, screens fill—and says nothing about whether the work that follows is faster. What a firm needs before it signs is a record of what the old workflow cost, kept long enough to be credible, and someone accountable for checking the model's output against what the old process produced. Kitces' ROI question is answerable only from that record; without it, the plug-in is an expense the firm cannot evaluate at renewal.

The supervision split turns concrete at exactly this point. A tool that reads across the stack also writes across it, and the firms that can log what the model touched, name who reviews it, and produce a client file that survives an exam will be the ones banking the time. Firms that cannot will pay the usage fee and run the manual check anyway.

Anthropic's advisor push is still being priced from the buy side too—the AI firm has asked wealth managers to submit pricing and service proposals. Advisory firms are the customer and the candidate at once, and a plug-in that touches the whole stack is the first product where both roles meet on a single invoice. The $240 seat was easy to approve because it was one line. This one will not be, so the number to write down is what the workflow costs today.