Anthropic rents the advisor's last mile, and rents get repriced

Anthropic's Claude for Financial Advisors arrived to enthusiasm from the vendor side of advisor technology and a narrower question from the analysts who grade that software for a living. Alois Pirker, the longtime industry analyst who now runs his own consultancy, called the launch positive for the industry and the outreach to practitioners ahead of it a good sign, then named the wall he expects Claude to hit: the same specificity problem that caught Salesforce, getting to what financial advisors actually need.

The precedent deserves the weight he gives it: nearly a decade old, and it points somewhere inconvenient for the launch narrative. Salesforce for Financial Services reached general availability a decade ago, later renamed Salesforce Financial Services Cloud, and the customization advisors wanted arrived through specialist firms like Practifi, the advisor CRM built on top of Salesforce for practices needing more tailoring than the parent platform would ever ship. The horizontal layer won the install base; the vertical layer did the work the install base was paying for.

Pirker sized the problem with LPL Financial, which runs more than 30,000 advisors and, within that one firm, probably 15 different types of advisor, each with its own products, service model and workflow — so a product tuned for one of them tends to arrive at the others as a demo. LPL carries $819.1 billion in registered assets across 2.85 million accounts and nearly 42,000 employees, the surface area on which a single workflow product stops looking like a platform, and any vendor selling AI into that channel is shipping fifteen products and calling it one.

Will Trout, who directs securities and investments research at Datos Insights, took the critique somewhere more concrete: the Kitces statistic Anthropic cited at launch — a typical advisory practice spends about a sixth of its time in client meetings — is real and has been real for years, and what matters now is whether Claude moves it. His proof of concept is the connector list. Claude plugs into Schwab, Addepar and Orion, and if those integrations work, he argues, the conversation stops being about technology and becomes a margin story.

Trout is right that the connectors are the product, and the sequencing is the strongest thing about this launch: Anthropic shipped into the custodians and the portfolio systems before the workflows were fully specified, putting Claude in front of advisors while its specificity is still thin and, more to the point, generating the usage data from which specificity gets built. Salesforce did it in the other order — general availability first, adaptation second — and the specialists collected the difference.

Where the margin framing goes wrong is in whose margin it is. As this publication argued when Schwab bought its zero-revenue seat at the front of Anthropic's queue, that seat is worth whatever the plumbing underneath it is worth, and the plumbing here belongs to Schwab, Addepar and Orion, which sit between an advisor and the money and can reprice access whenever they choose.

Trout's caveat carries further than the launch coverage did. The sixth-of-the-calendar figure has been true for years, meaning the tools aimed at the other five sixths have not moved it, and a model that moves it does so by converting saved hours into more clients per advisor or fewer people per client — a conversion that happens inside the firm owning the client relationship, not at the layer selling the tool. That is also why platform builders sound so enthusiastic about a model launch: better models make their plumbing more valuable and their claim on the saving larger.

Orion is already behaving that way, wrapping BlackRock, Fidelity and Vanguard models inside its tax engine, which suggests where Orion believes the value sits: in the layer that delivers the model and the firm that owns the client. Anthropic shipping a connector into that engine is Anthropic volunteering to be the model.

The platforms are investing accordingly. In August, LPL recruited Wells Fargo's technology chief to run platform engineering, a bet we flagged as platform engineering becoming the next retention weapon. Custodians and TAMPs write the plumbing because the plumbing is what keeps advisors, a third-party model that plugs into it is a supplier, and suppliers get repriced — the trade Anthropic has made: speed of distribution now, pricing power later, if it ever arrives.

Which leaves the analyst debate over specificity looking like the wrong argument. Specificity is a roadmap problem; given usage, a competent vendor converges on the workflow, and Pirker's own timeline — Salesforce reaching general availability a decade before the vertical specialists owned the niche — suggests the convergence runs in years, not architecture. The question that outlasts the roadmap is who owns the interface, and on this launch the incumbents do. The scenario that flips it is the one Trout flagged: Anthropic is positioning this as a layer rather than a vendor competing with advisors, and if the model gets good enough that advisors stop caring which system it sits inside, the connector list becomes a formality and the reprice never comes. The specialist workflow vendors have the most to lose in the interim, since every time a general model reaches par through a custodian's integration, their differentiation thins.

What to watch is unglamorous and specific: whether the Schwab, Addepar and Orion connectors stay free. The first of those firms to invoice for access will set the price for every AI tool that wants the advisor relationship after it, and that invoice will say more about where the value sits than the launch itself did.