Many advisory firms are adopting artificial intelligence tools quickly, but the bigger issue isn't whether AI works — it's how those tools are being layered into already complex workflows. Done poorly, firms end up with what I would call a "Frankenstack."
The conversations we're having with advisory firms about AI have changed a lot in the past year. The questions used to be exploratory: "What can AI do for us?" Now they're operational. Firms want to know how AI should change the way they serve clients and coordinate work across a team. That's a much harder question, because it exposes a fork in the road that most firms don't see until they're already committed to one side of it.
In 2025, our product team sought better insight into what clients and prospects were telling our sales and success teams. Important stuff was getting said in conversations across the business, and we weren't capturing any of it. As I tried three or four platforms that promised to pull insights out of conversations automatically, they all seemed like reasonable bets at the time.
But none of them stuck.
Some were too expensive, and some didn't fit. The real issue, though, was that I had misdiagnosed the problem. I was trying to buy my way out of what was actually a people and process problem. What we needed was for the team to be gathering insights ourselves, with a structure for what to do with them once we had them. Instead, I was shopping for software, and it took me longer than I would like to admit to realize that was the wrong move.
I see a similar pattern playing out in advisory firms right now.
Firms add tools to patch specific annoyances. Someone buys a meeting summarizer. Another team picks up a prospecting assistant. An additional unit starts testing an automated tagging tool. Each purchase makes sense on its own, but collectively the stack gets noisier, data fragments across systems and advisors end up doing extra work just to feed tools that don't talk to each other. The firm becomes "AI-enabled" in the same way a house with six different smart speakers is a "smart home."
A better process? Firms find the places where time and insight are getting lost, and redesign those workflows with AI as part of the structure. This is uncomfortable, and while most firms default to the tool path, here are four steps to integrate AI so it provides more than a set of parts that don't add up to a functioning whole.
Fix Data Fragmentation Before Layering AI
Point solutions split your data, so notes end up in one place, tasks in another and client history in a third. The advisor has to piece the story together before every conversation, introducing labor that nobody budgets time for. Someone has to figure out which system is current, reconcile records and manage permissions across tools that weren't designed to work together. A few power users are getting real value while everyone else quietly sticks with the old way.
There's a quieter issue too. Firms tend to ask, "Which tasks can we automate?" and that leads to retrofitting. You're putting a new engine in an old car and wondering why it still handles the same. A more appropriate question to ask is, "Where are we actually losing time and insight, and why?" For most firms, that answer lives in the handoffs: the gap between advisor and associate, between what's said in a meeting and what actually gets done about it afterward.
Redesign One Workflow Before Scaling AI
Firms making real progress tend to start with one workflow rather than buying five tools. They look for something that touches both client experience and internal coordination. Meeting prep works well because it involves relationship context, documentation and follow-through all in the same place. The important thing is defining what "better" looks like in terms you can observe: faster prep, more documented next steps, more proactive outreach tied to real client context.
Keep it grounded. Advisors need to see AI saving time inside the tools they already use, not in some separate interface they have to remember to open. And operations teams need to be able to watch the outputs and correct mistakes early, because AI amplifies what you feed it. If your CRM has missing relationships or inconsistent activity logging, the outputs will reflect that. You don't need perfection to start, but you do need a plan.
Pressure-Test AI Outputs Before Trusting Them
Vendors, meanwhile, need to move away from letting AI marketing language do the thinking for them. Ask how outputs are generated, what data is used, how access is controlled. Can you actually explain to a client how a recommendation was produced? Too many vendors are leaning on the word "AI" as if it's a feature in itself. Advisory firms operate under fiduciary expectations, and "We don't really know how it works" is not where you want to be when it comes to anything touching client outcomes.
One more thing to watch as you scale: whether your team is generating rework to correct bad AI outputs. If that's the case, you haven't saved time. You've moved it.
Assess Operations to Support an Existing Structure
Looking back at my own tool-chasing phase, I wish someone had told me that my first step should be to figure out how the team should actually operate. Once we did that, AI became really useful, because it was supporting a structure that already made sense. That's a less satisfying answer than "Buy this tool and watch the magic happen," but it's the honest one.
I think a lot of advisory firms are in a similar spot right now. The tools are impressive and getting better fast. But the firms that get something real out of AI will be the ones that did the unglamorous work first: looking at how they serve clients and redesigning from there. Clients won't know or care which model you licensed. They will absolutely notice when you're more responsive, and when you follow through on what you said you would do.
Conor Curtis is the head of product Practifi, a customer relationship management platform for advisors.
© Touchpoint Markets, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.