Every RIA I talk to has now bought AI. Very few have hired it.
That sentence sounds like wordplay. It isn’t. It’s the cleanest way I’ve found to describe the gap between firms that will get real leverage from AI in the next three years and firms that will end up with a drawer full of subscriptions and a story about why none of them stuck.
I’ve been in more than thirty firm-owner conversations in the last 30 days. Every single one started with some version of “we’re evaluating AI.” Almost none started with “we’re hiring AI.” Those are completely different projects.
The buying mindset
When you buy a tool, you care about features, price, ease of setup, the quality of the demo, and the case studies the vendor put together. You make a decision based on the product. The product is the point.
A buying mindset works fine for software that lives in one workflow. It’s how RIAs have evaluated CRMs, planning tools, portfolio reporting, and document management for twenty years. It produces decent outcomes when the thing you’re buying is contained.
The buying mindset is failing in 2026 because AI is not contained. An AI agent that drafts a meeting summary touches your CRM, your custodial data, your planning records, your compliance file, and the email you send your client tomorrow morning. It is not a feature. It is a co-worker. And you don’t onboard a co-worker by reading a feature matrix.
The hiring mindset
When you hire a co-worker, you care about a different list of things. You think about:
- The role. What is this co-worker accountable for? What does “good” look like in week one, month three, year one?
- Onboarding. What systems do they need access to? What context do they need from the team? Who answers their questions?
- Handoffs. What does work look like coming in, and what does it look like going out? Who reviews? Who approves?
- Performance. How do you know if they’re doing the job? What’s the metric — hours reclaimed, errors caught, client satisfaction, something else?
- Fit with the team. Does this co-worker make your existing humans more effective, or do they create more work to manage?
That list is not the list a buying-mindset RIA brings to a vendor demo. It’s a hiring manager’s list. And it produces a fundamentally different evaluation.
Why this distinction matters now
The wealthtech AI market in 2026 is like opening the Yellow Pages in 1995.
We have analysts trying to help us find the right tools and the features that will grow firms. The problem is tool sprawl. Pull up the Kitces AdvisorTech Map. Pull up the Oasis AI WealthTech Map. There are 100+ AI-specific firms on Oasis alone, plus hundreds more AI features stamped across the Kitces map. Good luck meeting with all of them.
The maps are working hard. They’re also wrestling with where this market is going — same as the rest of us.
Michael Kitces recently noted that some AI platforms don’t fit cleanly on the AdvisorTech Map because they’re not traditional SaaS — they’re “tech-enabled service providers” that belong on his newer Services Map. He called the tension between technology providers and tech-enabled service providers “very real and a notable part of how the industry is evolving.”
That’s the right read of the moment. It’s also the same moment every RIA owner is in. The categories we used to evaluate software were built for self-contained tools. The most valuable AI products in 2026 don’t fit those categories. They run operations. They span workflows. They look more like a hire than a SaaS subscription.
Oasis took a different swing at the same question. They built a separate map for AI because they saw co-workers — software that actually runs operations — as a new category, not a feature stapled onto a point solution. Both are right that the categorization is in motion. And no map can hire for you.
If you walk into that landscape with a SaaS buying mindset, you will end up with three to five AI tools that do not talk to each other, no audit trail, no clear measure of what’s working, and a team that’s tired of evaluating new software.
If you walk into the same landscape with a hiring mindset, the question shifts. It’s no longer “which AI is best?” It’s “what role do I want filled, and what does this co-worker need to actually do the job in my firm?”
Three things change immediately when you ask it that way.
First, you start evaluating data access before features. A meeting prep AI that can’t see your custodial feed is not a meeting prep co-worker. It’s a meeting prep tool that needs a human to do the actual prep around it. That’s not a hire. That’s a piece of software that creates work.
Second, you start evaluating handoffs. Your existing team is doing the work today. The new co-worker needs to take work from them, hand work back, and stay inside the rules of how your firm operates. If a vendor can’t tell you what those handoffs look like, the AI is going to live in its own world and your team will route around it.
Third, you start evaluating accountability. When the AI is wrong — and it will be wrong sometimes — who’s responsible? If the answer is “the human who approved it,” then you need a record of what the AI did, what it saw, and why it produced what it produced. That is not a feature. That’s the foundation that lets you defend AI-generated work to a client, a compliance officer, or a regulator.
What “hiring AI” actually requires
Once you start evaluating with a hiring mindset, the architecture you need becomes obvious. You need:
- A connected and orchestrated data layer so every AI co-worker has the same view of the firm your humans have. This is not your typical “integration play.” This is an intelligent, true orchestration of all of your data.
- A coordination layer above individual workers so handoffs happen cleanly across workflows instead of bouncing between tools.
- An audit trail that captures what the AI saw, what it produced, and how a human reviewed it — for every workflow, every time.
- A measurable outcome model so you know if the workforce is performing.
Most AI products in wealth management today don’t ship with all four. They ship with one — usually the most visible one, the agent itself — and leave the rest as your problem.
That’s the difference between an AI vendor and an AI workforce platform. And it’s the difference between buying an AI tool and hiring an AI co-worker.
The 18-month outlook
Three years from now, the firms two years ahead of the curve are going to be unrecognizable from the firms that bought five tools and never integrated them. The advisor headcount will look the same. The capacity per advisor won’t. The operating margin won’t. The valuation won’t.
That divergence is happening right now. It started the moment the second generation of agentic AI products hit the market — the ones designed to coordinate across workflows, not just sit inside one. Every quarter you delay the operating-model decision, the gap compounds.
If your AI plan in 2026 still reads like a vendor shortlist, that’s the gap to close this quarter. Don’t book another demo. Redraft the plan as an org chart.
Then go hire the AI workforce that fits.
Want to see what hiring AI looks like inside a $1B+ RIA? Book a 30-minute conversation — no slides, no pitch. We’ll walk your stack and your roles, not ours.
Sam Sova is the Co-Founder and CEO of Subatomic. Before co-founding Subatomic, he spent 20+ years leading digital transformation at Fiserv, TIAA, AT&T, and Johnson Controls.