Expert Q&A: Moving Beyond AI Experimentation

Firms across wealth and asset management have spent the last two years experimenting with AI solutions in an effort to transform their businesses, or just keep up with their competitors. Some efforts have paid off, but largely, firms haven’t made it to the point where AI tools are part of “business as usual”.

F2 serves RIAs, banks and trust companies, asset managers and asset servicers, supporting a wide range of their business functions through technology, operations, and marketing consulting. Two of our data and AI leaders, Jerry Robert, who supports RIAs, banks and other wealth management firms, and Swati Gupta, who supports asset managers and alternative asset providers, shared their thoughts on where firms find themselves with AI tools at this moment and what they need to do next to utilize AI’s power in an effective manner.

Q: AI investment across financial services has accelerated rapidly, yet many firms still struggle to move beyond experimentation. Why?

JR: Lacking a vision for your AI use keeps you locked in experimentation paralysis. If firms are honest with themselves, many started their AI journey due to FOMO. And when that’s your main reason for implementing AI, you’ll always be stuck in the experimentation stage. AI is a means to an end goal; it’s not the end goal. Your use of it should drive toward a goal. The firms that are successful with AI are using it deliberately.

SG: Exactly. To build on what Jerry said; firms have quickly launched pilots but most aren’t looking at their end-to-end business processes to see where AI capabilities can move the needle. They need to address their fragmented data, decide who owns this initiative, define the new operating model mechanisms, and measure impact. As Jerry said, they should work backwards from the goal so that they can scale AI intentionally and maximize its impact.

Q: How much does AI readiness ultimately depend on the quality of a firm’s data architecture, governance, and operating model foundations?

SG: Data has long been a foundational prerequisite for technology solutions. That’s more valid now with AI. We must address our fragmented data once and for all, and answer questions like, is the data current, accurate and cataloged? Is there lineage on the data? Can AI be used responsibly and are the results explainable? All of this becomes embedded in the new operating model.

Many firms need a data governance overhaul. Agentic AI, which will be embedded across the organizations, requires more governance. How do you manage access around these agents? What are the guardrails around these agents that are autonomously working on some of these business problems? This introduces the need for an AI Operating model (the AIOM).

JR: Yes, a catchy saying for that is “AI is the rocket, but data is the fuel.” Your fancy rocket isn’t going to Mars without fuel. AI is a pattern recognizer. It isn’t conscious, so your AI can only be as smart as your data. That means your competitive advantage is how you organize it, own it, distribute it, and orchestrate it. We have a four-pillar strategy for AI projects here at F2: data foundation, data governance and regulation, architecture and context management, and observability. Without these four pillars, we're not standing any AI data projects, because we need these four pillars for them to operate.

Q: Where are firms encountering the biggest disconnect between AI ambition and practical execution? Where are organizations underestimating the amount of operational and technology modernization required to support scalable AI adoption?

JR: Great question. The biggest disconnect I see is the concept that AI can do it all and drive revenue. Think about above the line and below the line activities in firms. Below the line is a sweet spot for AI. It’s doing rote tasks, like notetaking and checking documents for errors. AI can commoditize that science and we’re finding more use cases right now. Firms are getting some traction there in their operations.

But, above the line is the true revenue play. That client touch is the art. That's where the advisor actually works with a client, sits down, has coffee, plays golf, interacts with the client, etc. Above the line use cases are harder to execute. We need a lot more to get there. So people who are jumping directly into above the line and trying to build a co-pilot for the advisor without doing some of the below the line activity, struggle. Go for the low hanging fruit: commoditize the science, automate the swivel chair approaches. Once you've learned and you've grown, then attack the above the line pieces.

SG: I would add that for AI initiatives, the ambition is there. Everybody wants to get the new tools, new shiny toy, and deploy it and then assume they’ll get the operational efficiency or maybe revenue generation, but no one's investing in the organizational transformation that needs to happen. It isn't just a tech play. It's really around the entire data, tech, processes, and most importantly, the people. How are you elevating the people to utilize these tools as part of their day-to-day workflow and redefine their work. That's a leadership play. It's not just deploying the tool, but really embedding and understanding the change to the business. And so the ambition is there, but the disconnect also exists.

In addition, firms must address AI literacy to meet those ambitions. Practically, does the leadership understand the power of AI they have? Do they know how to embed it within their people and workflows? It's a top-down approach, not a ground-up approach. That's another gap between ambition and realization.

Q: Many organizations are layering AI initiatives onto already complex enterprise platforms and workflows. What separates firms simplifying operations from those creating additional complexity?

JR: Well if you're taking existing problems and trying to solve them through an old lens, you’re adding complexity. AI is a new capability. Reframe your conversations. Step back and look at what you're trying to do. Don't go back in there to just tinker with problems in the same way you always have.

SG: Complex systems exist in every organization we work with. Whether it's in the asset servicing, asset owner space, asset management, or in wealth management and bank trust. These are old companies that have built their systems for decades. They’re flying a plane and trying to re-engineer it while flying. That essentially is the added complexity. It’s important to maintain parallel processes with humans and AI together until they figure out the workflow and operating model, because AI is just going to amplify the good or the bad. We have to break down the entire process and make sure the workflows can embed AI as part of that operating model.

I'm going to go back to my point: leadership must understand what that new operating model is so that they embed that AI in those systems. We've seen this play out in the world of technology so many times over the years, but this is more than just a technology play. It's really important that we address this from the leadership perspective.

Q: From a client experience perspective, how should firms think differently about AI transformation to ensure it drives measurable value rather than isolated innovation efforts?

JR: AI tools empower the advisor, or augment the advisor in the interactions. Immediately people started talking about replacing the advisor. That’s a huge leap. First empower the advisor by making more time for them to interact with the client. Clients want humans to deal with them. As AI becomes more and more powerful, the value of human interaction is going to get more powerful. Wealthy clients are willing to pay a little extra to get personal service from an advisor. ‍

SG: In asset servicing it's really understanding that there are many valuable workflows for the entire end-to-end client journey. I'm speaking to a client right now about how important their onboarding is because the sooner the client is onboarded, the sooner their servicing can get streamlined on any new platform. So firms should make sure that they're adding the value of AI into that end-to-end journey.

Q: AI discussions often focus heavily on automation and efficiency. How do you see AI reshaping the workforce and the way teams operate day to day across both wealth and asset management?

JR: Commoditizing the efficiencies is the play that most firms will get into, which means it's going to impact processes and thus impact humans. The question is, what do you do when you’ve replaced someone entering PDF forms? Workforce reduction is going to be a reality, but what do you do with the efficiencies you gain there? Are you going to redeploy them? If you don't redeploy them, you lose the value. We must deliberately take those efficiencies gained from the science and deploy them into the art of our business. I think the smarter firms will probably take these efficiencies obtained and redeploy them into client experience. For example, an analyst who's worked with PDFs,maybe there's an opportunity to train that person to interact more with the client.

SG: I agree with Jerry. It's reshaping the workforce. We’ll need decision makers to make decisions on what AI is going to revert back. Decision making requires domain depth. So why lose your workforce who knows the domain so well when you can enhance their capabilities to make better decisions with AI? It’s a loss if firms aren't re-engineering the work itself. Plus, there's a lot more oversight that will be needed. If in three years all the workflows have embedded AI, and we’re advanced and matured, it will still need oversight on top of it. There will be autonomous agents running on their own that will need a continuous human feedback loop because there's going to be drift in the model, drift in our capabilities, drift in the data that's actually getting to these AI agents. So there's a constant loop that will need humans to rethink the workflow with agents. And I am actually very positively inclined to the new capabilities and jobs that will be coming in the marketplace for those needs.

Q: Looking ahead, what do you believe will define successful AI-enabled business transformation efforts across financial services over the next several years?

SG: Think about a year ago, everybody was just trying to get into prompting and vibe coding, now everybody's really good at it. In my mind, efficiency success is going to be defined within the next two years. This is a huge transformational change. But I think the principles in the AI Operating model is essentially where success will be defined. It's in the people, process, technology, and data. Everything around that really gets amplified.

JR: Those firms who haven’t been thoughtful about how they implement AI are going to see a steep financial impact, cost wise. For example, there was a news item that about $100K annual salary analyst ran up token usage costs of $100K in one month. Firms will get surprised with this kind of stuff over the next two years and all of a sudden say, I thought this was going to be efficient for me, but I've spent more money doing this than I ever did with my own resources. That’s why we show you how to future proof yourself. It’s why we have this four pillar strategy, and why we talk about your data foundation and your governance and regulation.

Learn more about how F2 can help you implement AI solutions strategically.

‍