Conversational Money: What Really Changes When People Talk to Their Bank Instead of Tapping It
The screen is quietly dissolving into a sentence. After twenty years of designing buttons, I've had to ask what my job becomes when there's nothing left to tap.
Strahil Hadzhiev
AUTHOR
For twenty years, the button was a promise. A sentence is a leap of faith.
Here's something I never fully appreciated until it started disappearing: a button is a promise.
When I put a "Send $500 to Alex" button on a screen, I was doing something more than styling a rectangle. I was showing you, before you committed, exactly what was about to happen. The amount. The recipient. The finality. You could read it, hesitate, change your mind, and nothing had moved yet. The screen's quiet genius was that it let you see the consequence before you caused it. Twenty years of my craft lived in that gap — the safe half-second between intention and action, where a person decides whether to trust what they're about to do with their money.
Now watch what happens when the interface becomes a conversation.
You type, or say, "send five hundred to Alex." And there's no form. No layout. No visible edges. Just your words, and then the system doing something. The gap I spent a career designing — the space where you see before you leap — collapses. Language doesn't have affordances. It doesn't show you the boundaries of what's possible or the shape of what you just triggered. It's you, a sentence, and a machine that heard something and acted on it.
Microsoft, Stripe, and PayPal are already wiring payment rails straight into chat. Tell an assistant "buy me a train ticket to Berlin for tomorrow" and it finds the options, compares them, and completes the purchase. That's not a demo anymore; that's a product category. And it sounds effortless, which is exactly what worries the part of me that's spent two decades watching where people get hurt around money.
Because effortless and safe are not the same thing. The tap interface was full of friction, and a lot of that friction was load-bearing. The confirmation screen you had to look at. The button you had to consciously press. The form that showed you every field before you submitted it. We spent years calling that friction and trying to reduce it — and it turns out some of it was the safety rail, not the obstacle.
So the first thing I've had to relearn is this: when you take away the screen, you take away the button's promise. And my job in conversational money is to put that promise back — not as a rectangle, but woven into the words.
The blank chat box is the most intimidating screen in finance
Give someone a form and they know what to do. Give them a blinking cursor in a chat box and ask them to manage their money, and most people freeze.
This is the first problem nobody warns you about with conversational finance: discoverability. A screen shows you your options. A conversation hides them. There's no menu, no visible set of things you're allowed to ask, just an open box that could accept anything — which in practice means people don't know what they can say, so they say too little, or the wrong thing, or nothing at all. The design challenge isn't making the conversation smart. It's teaching a nervous human, with no visible map, what this thing can safely do for them.
Then comes the harder problem: ambiguity. Language is gloriously, dangerously vague. "Send it to Mike" — which Mike? How much is "it"? "Pay off the card" — all of it, the statement balance, the minimum? A screen forces precision by making you fill in the fields. A sentence lets you be sloppy, and sloppy is fine when you're texting a friend and catastrophic when you're moving rent money. So the design work moves to the moment right before action: the system has to reflect back what it understood and wait. "I found a ticket for eighty-nine euros — confirm?" That single confirmation step is the new button. It's where the promise gets rebuilt. Skip it to feel smooth and frictionless, and you've designed a machine that spends people's money on a guess.
And then there's the part almost everyone gets wrong: register. Talking to your money is intimate. It feels like a person. And when something feels like a person, a chatty, breezy tone sneaks in — the little jokes, the emoji, the "Got it! 🎉" And that tone is a landmine, because the person on the other end might be checking whether they can make rent. The gap between how casual the interface sounds and how grave the moment actually is — that gap is where trust dies. I've learned to design the system's voice the way you'd coach a good private banker: warm, yes, but never flippant about someone's fear. Never celebrating a transaction the user isn't sure they wanted.
Underneath all three problems is one shift that changes everything about my craft. When the interface is a conversation, the words are the interface. Every sentence the system says is a piece of UX. The explanation of why a payment was blocked. The plain-language reason a loan was declined and what to do about it. The undo option offered the instant something feels wrong, with a clear note that the money's on its way back. In a tap world, copy was the thing we polished last. In a conversational world, copy is the product. A badly written sentence isn't a cosmetic flaw anymore — it's a broken button.
When the interface is a conversation, the writing is the design
So what actually becomes valuable, when there's no screen to arrange?
The same thing that was always valuable, finally standing in the open. Judgment about where a human needs to be slowed down. The instinct for where trust forms and where it shatters. And now, specifically, the ability to design a dialogue — the defaults, the confirmations, the graceful recovery when the machine misunderstands — and to write it in words that carry weight without adding fear. Conversational finance turns writing into an act of design. Not marketing copy. Trust architecture, one sentence at a time.
Part of that craft is knowing when not to be conversational at all. This is the counterintuitive skill. The whole appeal of talking to your bank is fluidity — say it, it's done. But the best conversational design knows exactly which moments to break the flow and force a hard stop. A large transfer. A change to who can access an account. Anything irreversible. At those moments the right design isn't a smooth reply — it's a deliberate wall: a clear summary, an explicit confirmation, maybe a step that pulls the human fully out of the chat and into a screen that makes them look at what they're doing. Frictionless is a feature everywhere except the exact places where friction is the point. Twenty years in financial products is largely twenty years of learning to tell those places apart.
And this is where the fundamentals I keep coming back to matter more than any trend. Strip away the chat interface and the voice assistant and the payment-in-the-message, and what's left is the oldest question I know: will a person trust this with their money, and how do I make that trust safe? Conversation doesn't change the question. It removes the screen I used to answer it with and hands me a blank sentence instead. That's harder, not easier. It demands more understanding of the human, not less — because now there's no layout doing quiet reassurance in the background. There's just the words, and whether they were written by someone who understands what's at stake.
That's the work. Not "add a chatbot." Design the moment a person speaks to their money and has to believe it heard them correctly, will do exactly what they meant, and will let them take it back if it didn't. That takes someone who has spent a long time watching where people flinch when money is involved, and who can now put that knowledge into a sentence instead of a screen.
The button is disappearing. The promise it made can't. Someone has to design that promise back into the conversation — and that someone should be a person who knew what the button was really for in the first place.
Strahil Hadzhiev
AUTHOR
