AI has exploded in usage and prominence over the past four years, and most of that usage has revolved around two main competing companies: OpenAI (with its ChatGPT product) and Anthropic (with Claude). And although ChatGPT and Claude are perhaps best known for their general-purpose chatbot interfaces, they're arguably equally important for the foundational Large Language Models (LLMs) they provide, which serve as the back-end of a host of other AI-powered software tools – including much of the AI-enhanced software that's used by financial advisors. In other words, tools that we think of as 'AI software for advisors' like Hazel, Slant, and Jump don't literally build out their own in-house LLMs: They connect to OpenAI's and/or Anthropic's LLMs and sit on top of those models to perform advisor-specific use cases. Which makes AI advisor technology effectively an advisor-friendly layer on top of an underlying general-purpose AI model, e.g., adding user interface, pre-developed prompts, and storage of or integrations to existing data sources, as features to run specific functions or workflows rather than relying on an open-ended chatbox window as both ChatGPT and Claude use, and often integrating 'deterministic' layers like calculators to do math that the 'probabilistic' AI model can't be fully trusted with.
But the difference between a general-purpose tool like ChatGPT or Claude and an advisor-specific solution isn't just the advisor-friendly interface and pre-built workflow capabilities; it's also the pricing model. The business-level tiers of ChatGPT and Claude that don't feed user data into the LLMs for training purposes (which already makes them a virtual requirement to purchase for protecting the privacy of client data) tend to be priced based not just on flat monthly fees per user but on the actual amount of 'tokens' used. As a result, using AI for complex reasoning tasks and multilevel agentic functions rather than relatively simple queries results in much higher billing rates as advisors exceed the pre-allocated amount of session or weekly tokens those tiers get by default and must pay for additional tokens. Advisor-specific AI software, on the other hand, still tends to bill solely on a flat SaaS fee model, even though the providers themselves are paying the LLM providers based on usage. Which works for the technology providers because they can set their pricing to account for the reality that some users will use the technology less than others (leaving the lighter users effectively subsidizing the heavier users who account for most of the token consumption), and is popular for advisory firms who can more easily predict their ongoing technology costs and don't have to worry about higher usage and "surprise" token bills cutting into their bottom line.
In that light, it's interesting to see Anthropic's announcement at September 2026's FutureProof Festival that it is launching a new Claude for Financial Advisors 'plugin' – essentially a set of pre-built skills and connections to the data from other advisor technology platforms that runs directly from the Claude chatbox – which seemingly aims to not only do away with the advisor-specific interfaces of third party software, but also the flat per-user pricing in favor of usage-based fees as Anthropic clearly hopes that advisors using Claude as their main interface to agentically execute all their multi-system workflows will use far more tokens than the standard business-tier allotments.
Notably, the intention of Claude for Advisors (at least for the time being?) seems not to be to replace any one or more existing advisor technology tools, but to move data in and out of the tools that an advisor already uses and run various workflows (e.g., creating a pre-meeting brief and agenda, checking marketing materials against SEC requirements under the Marketing Rule, or reviewing a client's portfolio drift against its target allocations). The initial feature set consists of eight different pre-built 'skills' (essentially workflows for the AI to run) and 11 different 'connectors' (data feeds to and from existing advisor technology tools) including Addepar, BlackRock, Black Diamond, Envestnet Tamarac, iCapital, MoneyGuidePro, Orion, Vanguard, Wealthbox, Wealth.com, and Zocks. Additionally, Claude for Advisors has announced that it will eventually be able to pull in custodial data from Schwab, although that connection isn't yet live as of this writing.
The idea, then, is to eliminate the need for advisors to jump back and forth between different software platforms to perform tasks that require data from multiple sources (i.e., the so-called 'swivel chair' problem). Instead of, for example, logging into Tamarac for investment performance data, MoneyGuide for the latest planning projections, Zocks for notes from the last annual review meeting, and then formatting all that information together in a Word document for the upcoming client meeting agenda, Claude can do the whole thing based on a prompt of "run /pre-meeting" for the client, with agents connecting to each system to prepare each component of the meeting agenda. On the plus side, this means that, for advisors who use the tools that Claude for Advisors connects to, they won't need to adopt a whole new set of tools to have an AI-connected tech stack, which might make Claude attractive for firms that don't want to endure the hassle of switching all their client data over to an internally-integrated tool like Hazel or Jump. But on the downside, Claude's functionality is limited to the number of technology providers that agree to share data with it, and the list of providers that is not on Claude's list of 'connectors' is as notable as those who are: Per the most recent Kitces Research on Advisor Technology, neither the two highest-adopted financial planning providers (eMoney and RightCapital), nor the most popular tax planning provider (Holistiplan), nor any of the top advisor marketing providers (FMG Suite, AdvisorStream, and Snappy Kraken), nor other popular tools like Advyzon, Nitrogen, or Jump are currently connected to Claude for Advisors. Additionally, while Schwab is the largest custodian on the market, advisors won't be able to pull in data from other custodians like Fidelity, Pershing, TradePMR/Robinhood, Interactive Brokers, or Altruist (although many Altruist users were ostensibly more likely to use Hazel than Claude to begin with). While some of these providers might eventually agree to connect with Claude, others might understandably be hesitant to do so if they see Claude as a potential competitor (since sharing client and advisor usage data might give Claude ample training material to build its own replacement products for the tech providers that agree to connect to it!).
From the advisor perspective, one of the big questions going forward will be whether advisors will be willing to buy a tool that, in addition to their existing software costs, will cost at minimum an extra $400 per month (since Anthropic requires Claude for Advisors users to be on an Enterprise pricing tier with a minimum of 20 seats at $20 per month), plus token usage costs which per Anthropic's informational webinar can run from $70–$120 per user per month. In other words, a solo advisor could expect to spend in the range of $5,500–$6,500 per year, while a 20-employee firm could expect to spend $20,000–$35,000 per year on Claude, all in addition to their existing software licenses. Which they may ultimately decide is worth it: If Claude can create enough extra efficiency to serve just one additional client per advisor, it will have more than paid for itself. But it does represent a significant upfront cost that advisory firms would need to incur, with no guarantee of whether Claude for Financial Advisors will be able to deliver on the time savings or extra revenue it promises. (Notably, Claude is offering a one-time $4,800 credit against token usage costs for firms that sign up by September 30, which reduces the initial commitment especially for smaller RIAs that would still need to sign up for a 20-seat enterprise license – but it doesn't change the ongoing cost of using Claude.)
However, aside from the question of whether/where the time savings or extra revenue will manifest, what's also still unknown is how effectively Claude can really pull together data from multiple tools into useful outputs for advisors without any speed bumps (e.g., advisors having to manually upload data from a source that isn't accessible to Claude) or hallucinations. After all, one of the popular selling points of advisor-specific AI tools is precisely that those tools have developed everything from the housing or cleaning of data to the guardrails and pre-built human-in-the-loop workflows to ensure nothing goes awry, which advisors just typing directly into Claude would have to figure out or monitor for themselves. Further, off-the-shelf tools have professional developers who can refine their workflows and agents to minimize the number of tokens used (relative to what non-developer advisors might execute by just typing their individual prompts into Claude). In addition, we've already seen in the last decade how the promises of seamless data integration (originally via API) can crash into the reality of unstable connections, providers' reluctance to share data with potential competitors, and advisory firms' data itself too scattered and unstandardized to be distributed cleanly throughout different ecosystems. While Claude for Advisors' debut can no doubt be viewed more as an MVP and a starting point than a finished product (i.e., it will likely iterate further and perhaps build more refinements from expanded integrations to establishing hallucination guardrails), that status leaves plenty of questions about what shape it will eventually take – and whether it can really come from the outside to supplant the tools that were built for advisors' use to begin with.
This article first appeared on the Nerd’s Eye View at Kitces.com at https://kitc.es/advisortech-oct2026, and has been reprinted here with permission.
Ben Henry-Moreland
Ben Henry-Moreland is a Senior Financial Planning Nerd at Kitces.com, where he specializes in writing and speaking on financial planning topics including tax, practice management, and technology. He also co-authors the monthly Kitces #AdvisorTech column. Drawing from his experience as a financial planner and a solo advisory firm owner, Ben is passionate about fulfilling the site’s mission of making financial advicers better and more successful.
Michael Kitces
Michael Kitces is Head of Planning Strategy at Focus Partners Wealth, which provides an evidence-based approach to private wealth management for near- and current retirees, and Focus Partners Advisor Solutions, a turnkey wealth management services provider supporting thousands of independent financial advisors through the scaling phase of growth.
In addition, he is a co-founder of the XY Planning Network, AdvicePay, fpPathfinder, and New Planner Recruiting, the former Practitioner Editor of the Journal of Financial Planning, the host of the Financial Advisor Success podcast, and the publisher of the popular financial planning industry blog Nerd’s Eye View through his website Kitces.com, dedicated to advancing knowledge in financial planning. In 2010, Michael was recognized with one of the FPA’s “Heart of Financial Planning” awards for his dedication and work in advancing the profession.
The breakaway-focused minority investor has staked 15 RIAs since 2024 as private sponsors increasingly trade stakes
Osaic subsidiary continues to forge new routes to US high-net-worth and ultra-high-net-worth people.
New data shows veterans report stronger financial security, legacy confidence and career earnings momentum than non-veterans.
Capco's Matt Berkowitz tells InvestmentNews that aggregating client data is no longer enough.
The capital most families plan to pass on will likely arrive too late to change anything.
Growing wealth, evolving priorities, and untapped opportunities are reshaping how women approach long-term investing.
Knighthead Life entered the market with a competitive MYGA. A strong launch earned advisor confidence and paved the way for FIAs.
