[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"cms-topics":3,"$f03eUpT8sbuyrWUa8S1tzDsEAdnAW8VeYXv4wwhuvo18":37},[4,14,22,29],{"slug":5,"name":6,"intro":7,"introText":8,"seoTitle":9,"seoDescription":10,"essayCount":11,"indexable":12,"path":13},"trust-as-infrastructure","Trust as Infrastructure",null,"Signals, systems, networks, time, and resilience - and how they build on each other. The conventional wisdom on trust is wrong. This is the correction. Each article breaks down one assumption about authenticity, credibility, or signals... and rebuilds it from the ground up. Strategy for leaders who need credibility to actually hold.","Trust as Infrastructure — Andrew Marconi","The conventional wisdom on trust is wrong. This is the correction. Strategy for leaders who need credibility to actually hold.",12,true,"\u002Fwriting\u002Ftrust-as-infrastructure",{"slug":15,"name":16,"intro":7,"introText":17,"seoTitle":18,"seoDescription":19,"essayCount":20,"indexable":12,"path":21},"human-in-the-loop","Human in the Loop","AI won't replace your expertise. It will expose whether you had any. Personal essays from the professional middle: the judgment calls, identity questions, and uncomfortable moments that don't show up in the frameworks. For practitioners navigating the AI transition without a script.","Human in the Loop — Andrew Marconi","AI won't replace your expertise. It will expose whether you had any. For practitioners navigating the AI transition without a script.",5,"\u002Fwriting\u002Fhuman-in-the-loop",{"slug":23,"name":24,"intro":7,"introText":25,"seoTitle":26,"seoDescription":27,"essayCount":20,"indexable":12,"path":28},"front-of-house","Front of House","The gap between what you intend to signal and what your audience actually receives is where trust dies. Practical reads on the visible layer... what your clients, readers, and critics actually see when they look. Because the strategy is only as good as what shows up on the other side.","Front of House — Andrew Marconi","The gap between what you intend to signal and what your audience actually receives is where trust dies. Practical reads on the visible layer.","\u002Fwriting\u002Ffront-of-house",{"slug":30,"name":31,"intro":7,"introText":32,"seoTitle":33,"seoDescription":34,"essayCount":35,"indexable":12,"path":36},"explorations","Explorations","Not everything fits in a series. These are the pieces that didn't... things I wrote because they were interesting. Not everything needs a framework. Field observations and investigations into specific industries, technologies, and territories.","Explorations — Andrew Marconi","Not everything fits in a series. Field observations and investigations into specific industries, technologies, and territories.",3,"\u002Fwriting\u002Fexplorations",{"essay":38,"related":171},{"slug":39,"title":40,"subtitle":41,"excerpt":42,"publishAt":43,"topic":44,"heroId":45,"path":46,"body":47,"bodyText":168,"seoTitle":169,"seoDescription":42,"canonicalUrl":7,"dateUpdated":170},"the-typing-bubble-is-lying-to-your","The “Typing…” Bubble Is Lying to Your Customers","Why faking human warmth reads as manipulation, not empathy","You added a 'typing...' delay to make your chatbot feel human. Your customer just realized they're being deceived by a progress bar.","2026-06-04T12:05:06.808Z",{"slug":23,"name":24},"04a0fb7b-afe3-47b1-9852-0fbada8dd93d","\u002Fwriting\u002Ffront-of-house\u002Fthe-typing-bubble-is-lying-to-your",{"type":48,"children":49},"root",[50,58,63,68,73,80,85,90,95,100,105,111,116,121,126,132,137,142,148,153,158,163],{"type":51,"tag":52,"props":53,"children":54},"element","p",{},[55],{"type":56,"value":57},"text","There are three dots pulsing at the bottom of the chat window. The little animated ellipsis that means someone on the other end is typing. You wait. For a second you extend the small social courtesy of it: a person is composing a reply, choosing their words, and the polite thing is to let them.",{"type":51,"tag":52,"props":59,"children":60},{},[61],{"type":56,"value":62},"Except no one is typing. The dots are on a timer.",{"type":51,"tag":52,"props":64,"children":65},{},[66],{"type":56,"value":67},"A developer set the delay to somewhere between two and four seconds because a product manager read that instant replies “feel robotic” and a slower reply “feels more human.” The pause is not thought. It is a loading animation wearing a costume.",{"type":51,"tag":52,"props":69,"children":70},{},[71],{"type":56,"value":72},"And the moment you work that out, and you will work it out, because the bot answers a 600-word policy question in the same three seconds it took to “type” hello, something curdles. You weren’t impatient a minute ago. You are now. Not because the answer was slow. Because the interface lied to you about why.",{"type":51,"tag":74,"props":75,"children":77},"h2",{"id":76},"the-warmth-costume",[78],{"type":56,"value":79},"The warmth costume",{"type":51,"tag":52,"props":81,"children":82},{},[83],{"type":56,"value":84},"These costumes are never made by villains. Someone reads that instant replies “feel robotic.” Someone gives the assistant a friendly first name. Someone draws a round avatar with a small smile. Three seconds of fake typing tests better than zero, so in goes the timer. Everyone at the table tells themselves they’re designing empathy.",{"type":51,"tag":52,"props":86,"children":87},{},[88],{"type":56,"value":89},"What they’re designing is a disguise.",{"type":51,"tag":52,"props":91,"children":92},{},[93],{"type":56,"value":94},"The whole act is borrowed from human conversation and bolted onto a machine: the name, the face, the typing indicator, the engineered hesitation. Each element does the same job, which is to make you forget, for as long as possible, what you are actually talking to. The intent is warmth. The mechanism is concealment.",{"type":51,"tag":52,"props":96,"children":97},{},[98],{"type":56,"value":99},"The problem was never the automation. People are fine being served by a machine. Roughly two-thirds of them actively prefer to handle the simple stuff themselves, no human required. The problem is the costume. The typing bubble doesn’t add warmth. It adds a tell.",{"type":51,"tag":52,"props":101,"children":102},{},[103],{"type":56,"value":104},"I think it’s the most quietly dishonest pattern in modern software, precisely because the people who build it mean so well.",{"type":51,"tag":74,"props":106,"children":108},{"id":107},"betrayal-is-not-a-slow-reply",[109],{"type":56,"value":110},"Betrayal is not a slow reply",{"type":51,"tag":52,"props":112,"children":113},{},[114],{"type":56,"value":115},"Here is the part the warmth instinct gets exactly backwards. People do not punish bad service the way they punish discovered deception. They are different injuries, and they live in different parts of the brain.",{"type":51,"tag":52,"props":117,"children":118},{},[119],{"type":56,"value":120},"Researchers who study trust have a name for the second one: betrayal aversion. In controlled studies, the mere possibility of being knowingly betrayed drops people’s willingness to engage by about a third, and the reaction runs hotter than the actual stakes justify. Hidden betrayal, the kind you find out about after the fact, drives people to walk away at measurably higher rates than honest bad outcomes do. The math is lopsided on purpose. The fake delay costs the customer three seconds. The discovery that it was fake costs you the relationship. Your nervous system knows those aren’t the same size, even when your dashboard doesn’t.",{"type":51,"tag":52,"props":122,"children":123},{},[124],{"type":56,"value":125},"So the typing bubble runs a terrible trade. It spends real, durable credibility to buy a few seconds of simulated presence. And it spends that credibility on the one thing it can’t actually deliver.",{"type":51,"tag":74,"props":127,"children":129},{"id":128},"the-thing-the-bubble-cant-fake",[130],{"type":56,"value":131},"The thing the bubble can’t fake",{"type":51,"tag":52,"props":133,"children":134},{},[135],{"type":56,"value":136},"EY’s research on AI in service puts a number on the hollow center. Seventy percent of people expect an AI system to understand the emotional context of what they’re saying. Forty-two percent believe it actually can. That twenty-eight-point gap is the empathy demand nobody is meeting, and the typing bubble is built to paper directly over it. It simulates attention, which is the precise thing the system isn’t paying.",{"type":51,"tag":52,"props":138,"children":139},{},[140],{"type":56,"value":141},"That’s why it reads as manipulation rather than care. Sixty percent of people, asked to choose, will take an empathetic resolution over a fast one. The bubble offers the posture of empathy with none of the substance. It’s a mannequin in a lab coat. And when the friendly mask is also the thing standing between you and an actual person, the friendliness curdles fastest of all: more than four in ten people say chatbots frustrate them specifically when they become the obstacle to reaching a human. The better the disguise, the longer it takes to realize you need to escape it.",{"type":51,"tag":74,"props":143,"children":145},{"id":144},"what-presence-actually-looks-like",[146],{"type":56,"value":147},"What presence actually looks like",{"type":51,"tag":52,"props":149,"children":150},{},[151],{"type":56,"value":152},"For years, Tony Hsieh’s Zappos refused to build a customer-service chatbot, and said so out loud, on brand grounds. The reason they gave is the entire argument in one sentence. A chatbot, they said, “wouldn’t hear your dog bark in the background and ask what type of dog you have.” They let agents stay on the line for as long as the customer wanted. Two of them logged calls close to eleven hours and were celebrated for it, not written up for blowing their handle time.",{"type":51,"tag":52,"props":154,"children":155},{},[156],{"type":56,"value":157},"That is the opposite of the typing bubble. It is not a machine performing presence. It is a person who is actually there, catching the dog, the suitcase, the thing you didn’t say. (Hsieh died in 2020, and the company that carries the name has reportedly let some of this erode, which is its own quiet lesson about how fast a real signal decays once nobody is tending it.)",{"type":51,"tag":52,"props":159,"children":160},{},[161],{"type":56,"value":162},"The fix here is not warmer automation. It is automation that stops pretending. A bot that says it’s a bot, answers fast, and hands you to a human the instant the human is the point. No timer. No costume. No little smile drawn on the front of the machinery.",{"type":51,"tag":52,"props":164,"children":165},{},[166],{"type":56,"value":167},"So open your own support flow and watch it answer you. If something in there is pretending to be a person, it isn’t buying you warmth. It’s spending trust you don’t get back.","There are three dots pulsing at the bottom of the chat window. The little animated ellipsis that means someone on the other end is typing. You wait. For a second you extend the small social courtesy of it: a person is composing a reply, choosing their words, and the polite thing is to let them. Except no one is typing. The dots are on a timer. A developer set the delay to somewhere between two and four seconds because a product manager read that instant replies “feel robotic” and a slower reply “feels more human.” The pause is not thought. It is a loading animation wearing a costume. And the moment you work that out, and you will work it out, because the bot answers a 600-word policy question in the same three seconds it took to “type” hello, something curdles. You weren’t impatient a minute ago. You are now. Not because the answer was slow. Because the interface lied to you about why. The warmth costume These costumes are never made by villains. Someone reads that instant replies “feel robotic.” Someone gives the assistant a friendly first name. Someone draws a round avatar with a small smile. Three seconds of fake typing tests better than zero, so in goes the timer. Everyone at the table tells themselves they’re designing empathy. What they’re designing is a disguise. The whole act is borrowed from human conversation and bolted onto a machine: the name, the face, the typing indicator, the engineered hesitation. Each element does the same job, which is to make you forget, for as long as possible, what you are actually talking to. The intent is warmth. The mechanism is concealment. The problem was never the automation. People are fine being served by a machine. Roughly two-thirds of them actively prefer to handle the simple stuff themselves, no human required. The problem is the costume. The typing bubble doesn’t add warmth. It adds a tell. I think it’s the most quietly dishonest pattern in modern software, precisely because the people who build it mean so well. Betrayal is not a slow reply Here is the part the warmth instinct gets exactly backwards. People do not punish bad service the way they punish discovered deception. They are different injuries, and they live in different parts of the brain. Researchers who study trust have a name for the second one: betrayal aversion. In controlled studies, the mere possibility of being knowingly betrayed drops people’s willingness to engage by about a third, and the reaction runs hotter than the actual stakes justify. Hidden betrayal, the kind you find out about after the fact, drives people to walk away at measurably higher rates than honest bad outcomes do. The math is lopsided on purpose. The fake delay costs the customer three seconds. The discovery that it was fake costs you the relationship. Your nervous system knows those aren’t the same size, even when your dashboard doesn’t. So the typing bubble runs a terrible trade. It spends real, durable credibility to buy a few seconds of simulated presence. And it spends that credibility on the one thing it can’t actually deliver. The thing the bubble can’t fake EY’s research on AI in service puts a number on the hollow center. Seventy percent of people expect an AI system to understand the emotional context of what they’re saying. Forty-two percent believe it actually can. That twenty-eight-point gap is the empathy demand nobody is meeting, and the typing bubble is built to paper directly over it. It simulates attention, which is the precise thing the system isn’t paying. That’s why it reads as manipulation rather than care. Sixty percent of people, asked to choose, will take an empathetic resolution over a fast one. The bubble offers the posture of empathy with none of the substance. It’s a mannequin in a lab coat. And when the friendly mask is also the thing standing between you and an actual person, the friendliness curdles fastest of all: more than four in ten people say chatbots frustrate them specifically when they become the obstacle to reaching a human. The better the disguise, the longer it takes to realize you need to escape it. What presence actually looks like For years, Tony Hsieh’s Zappos refused to build a customer-service chatbot, and said so out loud, on brand grounds. The reason they gave is the entire argument in one sentence. A chatbot, they said, “wouldn’t hear your dog bark in the background and ask what type of dog you have.” They let agents stay on the line for as long as the customer wanted. Two of them logged calls close to eleven hours and were celebrated for it, not written up for blowing their handle time. That is the opposite of the typing bubble. It is not a machine performing presence. It is a person who is actually there, catching the dog, the suitcase, the thing you didn’t say. (Hsieh died in 2020, and the company that carries the name has reportedly let some of this erode, which is its own quiet lesson about how fast a real signal decays once nobody is tending it.) The fix here is not warmer automation. It is automation that stops pretending. A bot that says it’s a bot, answers fast, and hands you to a human the instant the human is the point. No timer. No costume. No little smile drawn on the front of the machinery. So open your own support flow and watch it answer you. If something in there is pretending to be a person, it isn’t buying you warmth. It’s spending trust you don’t get back.","The Typing Bubble Is Lying to Your Customers","2026-08-05T22:32:07.995Z",[172,181,190],{"slug":173,"title":174,"subtitle":175,"excerpt":176,"publishAt":177,"topic":178,"heroId":179,"path":180},"your-seo-formatting-triggers-inauthenticity","Your SEO Formatting Triggers Inauthenticity Aversion","Why the shape signals 'machine' before anyone reads a word","Audiences aren't reading your words. They're scanning your formatting, recognizing the algorithmic structure, and bouncing before paragraph two.","2026-05-19T12:16:00.728Z",{"slug":23,"name":24},"0e1023b5-b98a-4931-ad25-cdf36bbf63f2","\u002Fwriting\u002Ffront-of-house\u002Fyour-seo-formatting-triggers-inauthenticity",{"slug":182,"title":183,"subtitle":184,"excerpt":185,"publishAt":186,"topic":187,"heroId":188,"path":189},"you-cant-teleprompt-trust-at-live","You Can't Teleprompt Trust at Live Events","Why over-scripted events do the opposite of what you paid for","You flew executives across the country and scripted every word. You just spent six figures to simulate an AI. Here's what trust at live events actually requires.","2026-05-12T12:11:51.470Z",{"slug":23,"name":24},"ce75fbfd-643b-4e6f-9abc-f27c8175d10e","\u002Fwriting\u002Ffront-of-house\u002Fyou-cant-teleprompt-trust-at-live",{"slug":191,"title":192,"subtitle":193,"excerpt":194,"publishAt":195,"topic":196,"heroId":197,"path":198},"the-scam-of-seamlessness","The Scam of Seamlessness","How frictionless design accidentally created perfect conditions for AI slop","The decade-long obsession with seamlessness built the perfect environment for AI slop. Adding friction back is the new premium signal.","2026-05-05T12:03:46.826Z",{"slug":23,"name":24},"d013ff7e-479e-4b19-a783-db65afb93757","\u002Fwriting\u002Ffront-of-house\u002Fthe-scam-of-seamlessness"]