At Schwab’s IMPACT conference, artificial intelligence wasn’t a side conversation.
It was the conversation.
In a session packed with roughly 700 to 750 attendees, John O'Connell, founder and CEO of The Oasis Group, asked a simple question: Who’s using AI?
“Every hand went up,” he said.
Six months ago, that wouldn’t have happened. Today, AI adoption across wealth management is no longer theoretical – it’s operational. But according to O’Connell, widespread use doesn’t mean best practices are in place.
In this Action Q&A, O’Connell breaks down what’s working, what’s overhyped and where firms risk moving too fast.
Suleman Din: What AI technologies stood out to you on the conference floor? What feels sustainable – and what feels like a flash in the pan?
John O’Connell: The first takeaway is simply the pace of adoption. Everyone is using AI. That’s a dramatic shift from even half a year ago.
The second big theme is note takers. Most firms are already using one. We’re nearing market saturation there, which means the next phase becomes feature competition and consolidation. If you look at the AI map we produce at Oasis, there are 17 firms in the note-taker category. There won’t be 17 in three years – probably not even in 18 months.
Beyond note-taking, AI-driven prospecting is gaining traction. We recently sponsored a study on AI prospecting that generated significant interest. Larger firms are implementing these tools first because they have dedicated business development teams. Smaller firms that rely heavily on referrals are slower to adopt, but I expect a sharp uptick in 2026.
Operational efficiency is the other big theme. From quick financial analysis to document ingestion, firms are starting to wrap their arms around the productivity gains AI can deliver. Those use cases will continue to expand throughout 2026.
Din: Where are the gaps? What’s not ready yet?
O’Connell: AI agents aren’t really there yet – at least not in plug-and-play form. If firms want meaningful agent capability, many will have to build it themselves.
Take document ingestion for a full financial plan. There are emerging solutions that can ingest tax returns or estate documents through providers like Wealth.com. But there’s still no comprehensive solution for many other document types – insurance policies, complex wills, alternative investments with partial ownership.
Larger firms that want serious automation will likely combine buying some technology with building some internally. And then they’ll need an orchestration layer on top of that – because that doesn’t really exist yet.
Din: Is adoption getting ahead of best practices? Are firms chasing AI FOMO without tightening data hygiene?
O’Connell: Unfortunately, yes.
Your existing processes are not adequate for artificial intelligence. If you want an AI agent to analyze prospect holdings, trust documents or bank statements, you must have guardrails in place.
If you simply give an agent access to your SharePoint environment, it may have access to everything – including HR files, confidential client data or compensation data that it should never touch.
Many firms are moving too quickly without strengthening cybersecurity, process controls and data hygiene. Those foundational elements are boring – but without them, you can’t safely accelerate. We’re likely to see some mistakes before firms fully internalize that.
Din: What about data ownership? Are advisors aware of where their data is actually going?
O’Connell: Buyer beware – or buyer must be aware.
If you’re using the standard paid version of ChatGPT for $20 a month, your data can still be used to train the model. To opt out, you need a team or enterprise version, which is much more expensive. Many firms think they’re protected when they’re not.
Then there’s vendor consolidation risk. If your note taker is acquired, how do you retrieve your data? Do you own it? Can you extract it cleanly?
And there’s third-party risk. An incumbent vendor may roll out AI features, but that AI capability may be licensed from another third party. Your data isn’t just with Vendor A anymore – it may be flowing to a large language model provider or other external partner. Many firms aren’t conducting that level of due diligence.
I would argue that some firms don’t even fully know where their data resides today.
Din: We saw robo-advisor consolidation after the initial boom. Do you expect a similar shakeout among AI startups?
O’Connell: Absolutely. We’re going to see consolidation. That’s inevitable.
The key is not to freeze, but to be disciplined.
First, create an AI acceptable use policy. Define approved tools, approved use cases and prohibited uses. Spell out how client PII and firm data must be protected.
Second, redesign your processes. AI requires new workflows with a human in the loop for verification and oversight. Advisors remain accountable for the output. You need review mechanisms.
Third, identify the use case that will generate the most value for your firm. Then conduct proper vendor due diligence before you buy.
If you rush out to grab a flashy tool, you’ll own a tool – but you may not get meaningful results. And you may not be meeting your fiduciary responsibilities in the process.
AI adoption in wealth management is no longer optional – it’s ubiquitous. But according to John O’Connell, maturity has not yet caught up with enthusiasm.
Note takers are nearing saturation. Prospecting tools are gaining traction. Operational efficiency is accelerating.
Yet the real differentiator won’t be who adopts fastest. It will be who builds the strongest foundation – with governance, data control and process discipline – before scaling.
In AI, as in investing, speed without strategy rarely wins.