You know the sound.

A wooden door swinging open — that hollow, slightly too-loud creak that meant someone you knew had just come online. And the reverse: a door closing, decisive and final, which meant they were gone. AOL Instant Messenger used those two sounds as its entire presence system from 1997 onward, and if you grew up in the United States between roughly 1999 and 2007, they are filed somewhere in your brain with the same permanence as your childhood phone number or your first locker combination.

AIM shut down on December 15, 2017. Verizon, which by then owned AOL through its Oath media subsidiary, announced it two months before with a brief blog post. No ceremony. No final transmission. The servers just stopped accepting connections, and that was that.

Most people didn’t notice for days.

The Origin

AOL Instant Messenger launched publicly in May 1997, though it had existed in a more primitive form as part of the America Online client since 1996. The broader context matters: the late 1990s internet was a very different object than what came after. Broadband was not widespread. Most home connections were dial-up — 56k at best, which meant the connection tied up the phone line and cost by the hour in some plans. Email was slow. The web was mostly static. Real-time communication between two people who weren’t in the same room was not yet something ordinary people did as a matter of course.

AIM changed that. The core product was simple: a buddy list, a chat window, and a presence indicator that told you whether the person was online, away, or offline. You could type a message and have it appear on their screen within seconds. This sounds trivial now. In 1997, it was genuinely new.

What made it stick was the away message.

The Away Message

This is the part that gets underwritten in retrospectives about AIM, and it shouldn’t be.

The away message was a field you filled in when you stepped away from your computer. In theory it was a functional notice — “at dinner, back at 8.” In practice it became something much stranger and more significant: the first widespread format for public, low-stakes self-expression on the internet.

By 2001, teenagers were using away messages the way people would later use Twitter bios and Instagram captions. You put in a song lyric that described how you were feeling without having to say directly that you were feeling it. You posted an in-joke that only certain people on your buddy list would understand. You wrote something cryptic aimed at one specific person who you knew would read it. The away message was a proto-status update, a micro-blog, a passive-aggressive communication channel, and a form of performance all at once — and it existed almost a decade before any of those terms entered the cultural vocabulary.

The profile was the same. AIM let you write a profile — essentially a public bio — that anyone could read before or during a conversation. Fourteen-year-olds in 2002 were crafting personal brands on AIM profiles without knowing that’s what they were doing. Favorite movies, favorite quotes, which friend group they belonged to, what they wanted people to know about them. The format was completely open-ended and deeply personal in a way that later, more designed social networks struggled to replicate.

What It Was at Its Peak

By 2001, AIM had 53 million users and controlled roughly 52 percent of the instant messaging market. That is not a niche product. That is infrastructure. In American middle schools and high schools of that era, not having AIM was roughly equivalent to not having a phone number — it meant people couldn’t reach you in the way that had become normal.

The culture that grew up around it was specific and elaborate.

Buddy icons were tiny, 48x48 pixel images that appeared next to your screen name — essentially the first avatar system most people ever interacted with. People traded them, made them, argued about them. The best ones were animated. There were entire websites dedicated to nothing but hosting buddy icons.

Screen names were an art form. You had one identity online — your screen name — and people thought about them carefully. They combined words in unexpected ways, included numbers that meant something to them, gestured at interests or personality traits. Your screen name was your handle on the internet in a way that email addresses never quite were, because email felt formal and AIM did not.

The sounds. Beyond the door creaking open and closed, AIM had a library of notification sounds — messages arriving, alerts firing — that became ambient noise in any house with a teenager between 1998 and 2005. Parents knew what it meant when they heard the door sound from the other room. Friends communicated in those sounds without words: a door sound meant someone was now available. A series of rapid message alerts meant something was happening.

And then there was SmarterChild.

SmarterChild

In 2001, a company called ActiveBuddy launched a chatbot on AIM called SmarterChild. You added it to your buddy list like any other contact and then chatted with it.

SmarterChild could answer questions, tell jokes, play games, look up sports scores, check movie times, and engage in what passed for conversation. At its peak it had over 30 million users across AIM and MSN Messenger — which made it, by some measures, one of the most-used conversational AI systems in the world at the time. People spent hours talking to it. They tried to break it, befriend it, confuse it. They asked it questions they wouldn’t ask anyone else. An automated chatbot with a personality, living in a messaging client, having millions of conversations a day.

This was 2001. ChatGPT launched 21 years later and is described as a new category of technology.

ActiveBuddy sold to Colloquis, which was acquired by Microsoft in 2006. SmarterChild’s original code formed the basis of what eventually became early versions of Microsoft’s enterprise chatbot infrastructure. The lineage from a novelty AIM bot in 2001 to modern conversational AI is not as indirect as it looks.

The Decline

AIM didn’t lose to a single competitor. It lost to a structural shift it couldn’t adapt to.

The first blow was the mobile phone. Through the mid-2000s, SMS text messaging — which had been a slow-moving feature of mobile phones — became cheap and then free. Teenagers who had been chatting on AIM from desktop computers now had a way to message their friends from anywhere, without needing to be home, without needing to log on. The presence indicator that made AIM feel magical — knowing whether your friend was online — became less relevant when you could just text them and expect a response eventually.

The second blow was Facebook. Facebook launched its chat feature in 2008. By that point Facebook had already absorbed most of what AIM’s social layer had been doing — the profile, the status update, the semi-public broadcast to a defined group of people you knew. When Facebook added real-time messaging, it wasn’t just competing with AIM; it was offering everything AIM offered inside a platform that had already become where people spent their time online.

AOL knew it had a problem. The company tried to pivot AIM repeatedly — toward video chat, toward social features, toward mobile. None of it worked. The product that had once felt like the future of communication now felt like a holdover from the era of dial-up. By 2011, AIM’s market share had fallen from 52 percent to under 1 percent. By 2017, Oath — the Verizon subsidiary that owned AOL, Yahoo, and a collection of other legacy internet assets — decided the cost of maintaining the servers was no longer justified.

The shutdown notice, published in October 2017, was 180 words long.

The Last Day

On December 15, 2017, AIM went offline.

The response online was a wave of genuine grief — not for a product, exactly, but for a period. People posted screenshots of old buddy lists. They wrote about conversations they’d had on AIM in high school that they still thought about. They described the specific texture of that form of communication: the way you could see when someone was typing, the way the away message let you signal something to everyone without saying it to anyone, the way the buddy list was a map of your social world at a given moment in time.

What they were mourning was not a piece of software. They were mourning a version of themselves that existed before their communication was mediated by platforms with algorithmic feeds, engagement metrics, and advertising infrastructure. AIM felt intimate in a way that later social networks don’t, because it had no audience beyond the people you explicitly put on your buddy list. There was no viral spread, no public performance, no follower count. Just a list of screen names and a blinking cursor.

That feeling — of talking to a specific person in something that felt like a private space — is the thing that disappeared when AIM did. The technical function of sending a text message to someone has never been more available. The feeling of it has never been harder to find.

What It Left Behind

Every major messaging platform running today is working from a template that AIM established.

The presence indicator — the green dot that tells you whether someone is active on iMessage, WhatsApp, Instagram, Slack — is the door sound translated into a pixel. The status update is the away message at scale. The avatar is the buddy icon with more resolution. The concept of a contact list that maps your social world and shows you who’s available in real time: AIM’s design, running on infrastructure with a thousand times more users.

The platforms are different. The scale is different. The business models are radically different. But the fundamental vocabulary of digital communication — the idea that you should be able to see who’s available, signal your own status, and send a message that arrives immediately — was worked out in the AIM client between 1997 and 2007, by people who mostly had no idea what they were building.

AOL spent $407 million acquiring ICQ in 1998 — AIM’s main early competitor, and arguably the first true consumer instant messenger — which gave it a monopoly on the instant messaging market at the exact moment instant messaging was becoming important. That monopoly should have been a lasting structural advantage. Instead it became a position to defend rather than extend, and the defense eventually failed.

The mistake wasn’t technical. It was a failure to understand that AIM’s real product was not the messaging client. The real product was the social graph — the buddy list, the relationships, the web of people who had added each other and developed habits of communication. When Facebook built a social graph with more features and locked it behind a walled garden that AIM couldn’t penetrate, the buddy list became less useful every day a friend migrated. Network effects that had once worked for AIM began working against it.

By the time AIM shut down, its user base had already left. The shutdown was just paperwork.

The door sound was recorded specifically for AIM. Nobody seems to know exactly who made it or when. It wasn’t a sampled sound effect from a library — someone sat down and recorded a door opening and closing and decided that was how it would feel to know a friend had come online.

Fifty-three million people heard that sound and understood exactly what it meant.

When the servers went down on December 15, 2017, nobody played it. The door just stopped opening.