Tech providers are starting to push agentic AI, but do we really just need better UX and automation?

At times over the past two years, it's felt like technology development has gone into hyperdrive mode, as AI tools first burst into the mainstream with chatbot-style Large Language Models (LLMs) such as ChatGPT, and now increasingly are becoming embedded into much of financial advisors' everyday technology. The most prominent example is in the client meeting notes category, which went from zero to (at last count) 12 different advisor-specific tools on the Kitces AdvisorTech Map in just two years, but AI has more quietly become embedded into other tools as well, from planning tools like FP Alpha's AI-powered NextGen Tax Insights to AI-embedded CRMs like Practifi to all-in-one platforms like Orion launching an AI-powered meeting prep tool.

With AI spreading through the industry so fast, it's been natural to occasionally take a step back and wonder whether it's all really necessary, and whether AI will really become a permanent feature of everyday technology, or whether it's just a hype bubble that will burst at some point and all of those AI tools will prove too expensive to maintain at a price that people will pay for them once the flood of venture capital funding that's kept them afloat starts to dry up. At this point, however, as advisors appear to be becoming more comfortable with the use of AI, and as AI becomes more embedded in the tools advisors use every day, it seems likely that even if there is a pullback at some point, AI will still have some kind of presence in the advisor technology landscape – even if, given the still-rapid pace of change, we don't know exactly what that presence will look like yet.

According to many tech vendors and investors, the next frontier in AI is what's being referred to as "agentic AI". The definition of agentic AI seems to differ depending on who's being asked, but the way it's often described is as a tool that can take a single request from a user and perform, unprompted, a sequence of tasks that accomplish that request – for instance, booking a flight, responding to a customer service inquiry, or researching and emailing a prospective client. Also according to those same sources, agentic AI is either just a few months from widespread deployment or else it's already here, ready to drastically reduce staffing and overhead costs by eliminating the need for everything from administrative and customer service staff to researchers to software programmers to web designers.

In a financial advisory context, there are essentially two ways that agentic AI could theoretically be employed. But unfortunately, neither of them really seem to make sense for advisors and their businesses anytime soon.

For the first use, there are tasks that are already generally the job of the financial advisor: Finding and meeting with prospective clients, creating and analyzing financial plans for clients, and meeting with clients and giving recommendations. And it's hard to imagine any of those being fully replaced by agentic AI. There's conceivably a role for AI in prospect research (and several AI-powered prospecting tools like FINNY, AIdentified, and Wealthawk, have already emerged in recent years), but the advisor isn't going to outsource prospect meetings themselves to an AI agent; at some point, the advisor still has to show up, build a relationship, demonstrate their trustworthiness, and persuade a prospect to hire them. Likewise, few advisors are going to outsource client meetings with AI. That leaves financial plan creation and analysis, which is often time-consuming and inefficient, but in a fiduciary business where the advisor is professionally accountable (and financially liable) for any advice they give, it's hard to imagine many advisors outsourcing what is usually the primary source of data to back their recommendations to a technology tool that can't be held accountable for any inaccuracies in its output. Which means at best, perhaps the AI tools ‘just' gather the data, input it to the software, and come up with a list of "potential" recommendations for the advisor to consider… a non-trivial value, but likely saving no more than a few hours per new financial plan (when most advisors take on only 6–12 new clients per year).

The other use cases are in areas of the advisory business that are not generally handled by the advisor: Client service, trading operations, compliance, marketing, etc. And while there are places where AI can assist humans who work in all these areas, what that often comes down to is automating certain sets of tasks – e.g., workflows for opening client accounts or money movements, rebalancing client accounts, or going through a pre-meeting checklist. And given how process-based most advisory businesses are operationally, these are not the sorts of tasks that even actually require AI to automate, since they usually follow the same sequence of steps each and every time they're performed… making them more conducive to good 'old-fashioned' business process automation than agentic AI.

For example, imagine that after every client meeting an advisor follows the same sequence of steps: logging their client meeting notes into their CRM; kicking off follow-up tasks to members of their team for each action item; and sending a meeting recap email to the client. Although each of those steps can be individually helped by AI (e.g., meeting note software to convert the meeting transcript into a summary, a chatbot in the CRM to assign tasks without manually creating each one, and an AI writing aid to create the first draft of the email), the whole process doesn't need an agentic AI to complete it, because there's no decision-making involved: It can be achieved much more easily with a simple automation running between the advisor's AI notetaker and their CRM software that runs the exact same sequence for every client meeting. And if the advisor ever needs to change that workflow, it's much easier to do so in the context of an automation (where they can simply add, remove, or edit steps in the predefined sequence) than it is with an AI tool that must be re-trained with a series of prompts on how to do its job differently.

In other words, while it's easy for tech platforms and their users to be drawn in by the shiny new object of agentic AI, it's worth wondering for every function that AI purports to perform whether it's really necessary for an AI agent to do it – that is, does it involve real decision making on behalf of the technology and applying the wants and needs of the user to the circumstances at hand? – or if the AI is really just being used to patch over a lack of meaningful integrations or automation capabilities or poorly designed UX that makes it harder than it ‘should' be for an advisor to accomplish a task or workflow. Because as it stands in the advisory business, the actual functions that can or should be handled by an AI agent seem fairly limited – either because they're already core parts of what the advisor does, they shouldn't be done by AI because they're central to the advisor's fiduciary duty, or they're really just automations that could be done more easily with a better user interface in lieu of AI altogether?

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 is the Chief Financial Planning Nerd at Kitces.com, dedicated to advancing knowledge in financial planning and helping to make financial advisors better and more successful. In addition, he is the Head of Planning Strategy at Focus Partners Wealth, the co-founder of the XY Planning Network, AdvicePay, New Planner Recruiting, fpPathfinder, and FA BeanCounters, 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.