Three Conversations That Changed How I Think About Selling AI

I've had hundreds of conversations about AI in financial services. Pitches, demos, follow-ups, the whole thing. Most of them blend together. You get into a rhythm. You know the objections before they come. You know which slide makes people lean forward and which one makes them check their phone.

But three conversations broke the pattern. They happened in different cities, with different people, about different problems. And they fundamentally changed how I think about what I do.

I'm going to tell you about all three. The names and some details are changed, but the substance is exact. These are things people actually said to me, in rooms where they thought no one was taking notes.

I was.

Milan: "I don't need better AI. I need fewer vendors."

This was late on a Tuesday afternoon in a glass-walled conference room near Porta Nuova. The kind of room where you can see the whole Milan skyline but nobody ever looks at it because the conversation is too tense.

I was meeting with the CTO of a mid-size Italian bank. Not one of the giants, but serious. Several hundred billion in assets under management. A technology team of maybe 200 people. He had agreed to the meeting because a mutual contact had made the introduction, and he was curious about what we were building.

I was fifteen minutes into the pitch. It was going well, I thought. I was showing how our platform could improve their advisory workflow, how the AI models could surface insights their wealth managers weren't catching, how the integration was clean and the compliance layer was built in.

He held up his hand.

"Let me stop you. I'm sure your product is good. I've seen a lot of good products this year. But I need to tell you something that might save us both some time."

He pulled up a spreadsheet on his laptop and turned it toward me. It was a list of vendors. I counted quickly. There were over 40 technology vendors with active contracts at this bank. Forty. Some were legacy infrastructure, sure. But at least fifteen had been added in the last two years, most of them AI-related.

"I don't need better AI. I need fewer vendors. Every new tool is another integration, another contract, another security review, another team that needs to learn something new. My people are drowning. They spend more time managing vendor relationships than doing actual technology work."

He wasn't angry. He was exhausted. And he was right.

I sat with that for a long time after the meeting. I walked back to my hotel through the Navigli district, and I kept thinking about that spreadsheet. Forty vendors. Each one had probably pitched exactly like I had. Each one had a great product. Each one was solving a real problem. And collectively, they were creating a bigger problem than any of them solved.

The lesson: Banks are not suffering from a lack of technology. They're suffering from vendor fatigue. The real competitive advantage isn't having the best AI model. It's being the vendor that makes the other vendors unnecessary. Consolidation is the product. Simplicity is the product. If you walk into a bank and your pitch adds complexity to an already overstretched technology team, you've already lost. You just don't know it yet.

After that meeting, I started every pitch differently. Instead of leading with features, I started asking: how many vendors are you working with right now? What would it mean to reduce that number by three or four? The conversations changed immediately. People leaned in. Because nobody else was asking that question.

Doha: "Will your AI make my clients trust me less?"

A few months later I was in Doha for a series of meetings with wealth management firms. The Gulf states are fascinating for AI adoption because the money is there, the ambition is there, and the regulatory environment is more flexible than in Europe. But the client relationships are different from anything you see in Milan or Zurich.

I was meeting with a senior wealth manager at a private bank. Not the CTO, not the innovation team. An actual relationship manager who handled a small number of ultra-high-net-worth families. He had been in the business for over twenty years. His clients' parents had been his clients. This was generational trust.

We were talking about how AI could help him prepare for client meetings. Portfolio analysis, market summaries, risk scenarios. The kind of work that takes hours of preparation and can be compressed into minutes with the right tools. He was nodding along. It made sense to him.

Then he asked the question that stopped me cold.

"If my client finds out that the insights I'm sharing with him came from an algorithm and not from my own analysis, will he trust me more, or less?"

He wasn't being rhetorical. He genuinely wanted to know. And his concern was specific. His clients chose him because of him. His judgment, his experience, his personal attention. They were paying a premium not for performance, which they could get from an index fund, but for the relationship. For the feeling that someone they trusted was watching their money.

"My clients don't want to hear that a machine helped me. They want to hear that I stayed up until midnight reading the quarterly report myself. That's what trust looks like in my world."

This was a completely different objection from anything I'd heard in Europe. In Milan or London, the pushback was usually about compliance, integration, cost. Technical objections. This was an existential objection. He was asking: does your product undermine the very thing I sell?

The lesson: In private banking, trust is the product. Not returns, not alpha, not insights. Trust. And trust is deeply human. It's built over dinners and phone calls and showing up at funerals. If AI threatens that human layer, even just the perception of it, the technology is worthless. It doesn't matter how good the model is.

This changed how I position AI in wealth management entirely. The framing shifted from "AI that gives you better insights" to "AI that gives you more time with your clients." The technology fades into the background. The human relationship comes to the front. The wealth manager doesn't become more automated. He becomes more present. He's the one who stays up until midnight, but now he spends that time talking to his client instead of building a spreadsheet.

The distinction sounds subtle. It isn't. It's the difference between a pitch that gets a polite "interesting" and one that gets a second meeting.

London: "My job is to run pilots, not ship products"

This one was the hardest to hear. Because it wasn't about the market or the client. It was about the internal machinery of the organizations I was trying to sell to.

I was in London, at a large bank. One of the names everyone knows. I was meeting with their Chief Innovation Officer, a title that's become standard at European banks over the last few years. He had a team of about thirty people, a dedicated innovation lab, a portfolio of partnerships with fintech companies. On paper, this was the perfect customer.

The meeting was going well. His team was technically sharp. They understood the product immediately. They saw the value. They wanted to move forward. And then, over coffee after the formal meeting, the CIO said something that I've thought about almost every day since.

"I'm going to be honest with you. My job is to run pilots. That's what I'm measured on. How many pilots am I running? How many startups am I talking to? How many innovation reports am I presenting to the board? Nobody has ever asked me to ship a product into production. That's someone else's problem. And that someone else doesn't report to me."

He wasn't being cynical. He was being accurate. The bank had created an innovation function that was structurally disconnected from the teams that actually built and deployed technology. The innovation team found interesting things. Then they handed them to IT, who had their own priorities, their own budget cycles, their own roadmap. And the interesting things died in the handoff.

He told me that in four years, his team had launched over twenty pilots. Two had made it to production. This is the pilot trap at its most honest. Two out of twenty. And both of those had succeeded only because a specific executive had personally championed them through the organization, overriding the normal process.

"The technology is never the problem. The org chart is the problem."

The lesson: The organizational structure of a bank determines what gets adopted, not the quality of the technology. If innovation and implementation sit in different silos with different incentives, nothing crosses the gap. You can have the best product in the world, and it will die in a pilot. The question isn't "is this a good product?" The question is "does this organization have a pathway from pilot to production?" If the answer is no, you're wasting your time and theirs.

After London, I started qualifying opportunities differently. Before I even pitch the product, I try to understand the internal structure. Who owns the decision to go from pilot to production? Is that person in the room? If not, can we get them in the room? If the answer is "the innovation team decides to pilot and then IT decides whether to deploy," I know exactly what's going to happen. We'll run a beautiful pilot. The results will be impressive. And nothing will come of it.

I'd rather walk away early than waste six months on a pilot that was never going to become a product.

What these three lessons have in common

Milan taught me that the problem isn't technology. Doha taught me that the problem isn't insight. London taught me that the problem isn't innovation.

The problem is always human. Always organizational. Always about incentives, relationships, and trust.

The AI industry, especially in Silicon Valley, has a blind spot the size of a continent. We build better and better models. We optimize for benchmarks. We talk about capabilities. And we assume that if the technology is good enough, adoption will follow.

It won't. Not in financial services. Not in Europe. Not in the Gulf. Probably not anywhere that the stakes are high enough that people can't afford to get it wrong.

Adoption follows trust. Trust follows simplicity. Simplicity follows a deep understanding of how the customer's world actually works.

That's what these three conversations taught me. And that's what shapes how we work at Streetbeat now. We don't lead with the model. We don't lead with the features. We lead with the question: what is actually hard about your day? And then we shut up and listen.

Because the answer is almost never what you expect. It's never "I need a better algorithm." It's "I need to spend less time on things that don't matter so I can spend more time on things that do." It's "I need to trust that this won't blow up in my face." It's "I need someone who understands that my world is more complicated than a demo suggests."

The best technology disappears. You don't notice it. It just makes the important things easier and the unimportant things go away. That's what I'm trying to build toward. Not AI that impresses people in a meeting room, but AI that a wealth manager in Doha forgets is even there, because he's too busy talking to his client. That is what the bank of 2030 actually looks like.

Three cities. Three conversations. Three lessons I carry into every meeting now.

The fourth conversation could be with you. I'm listening.