There is a small piece of software that has quietly followed you around the internet for most of your adult life. It is the little tab that says "Chat with us." It waits in the corner of your bank's website, your airline's app, your phone carrier's help page. You have almost certainly typed into it. You have probably never wondered who built the thing on the other side.
A lot of the time, the answer is LivePerson. The New York company has been in the business of brand-consumer conversation since 1995, which makes it older than Google, older than the smartphone, and old enough that its founder, Robert LoCascio, was placing live human agents behind chat windows when most of the web was still under construction. What started as a novelty - a real person, answering in real time - became an industry. LivePerson helped invent it.
The company's product today is called the Conversational Cloud, and its scale is the kind that hides in plain sight. It runs close to a billion conversational interactions a month for more than 18,000 customers across 21-plus industries. You will not see its logo on a billboard. You will see its work every time you message a company instead of calling it.
01 / WHAT IT DOESThe switchboard for a messaging world
Strip away the jargon and LivePerson does one thing: it sits between a brand and its customers and manages the conversation. When you text your bank, message an airline on WhatsApp, or tap "Message" in an app, the words have to go somewhere, be understood, be routed, and be answered - sometimes by a human, increasingly by an AI, often by both handing off to each other mid-sentence. The Conversational Cloud is the machinery that makes that handoff invisible.
It plugs into the channels people actually use: SMS, WhatsApp, Apple Messages, Google's RCS, Facebook Messenger, LINE, WeChat, and the native chat inside a company's own website and app. On top of that pipework sits the interesting part - automation. Brands build AI agents that can answer routine questions, and a unified workspace lets human agents review or approve what the AI wants to say before it goes out. The pitch to a supervisor is control; the pitch to a customer is speed.
02 / THE PROBLEMTurning a cost center into a place people buy things
Most companies think of customer service the way they think of plumbing: necessary, invisible when it works, expensive when it breaks. It shows up on the balance sheet as a cost to be minimized. LivePerson's core argument, repeated for years, is that this framing is a mistake. The support conversation, it says, is one of the few moments a customer is actively paying attention - and that attention can be sold to.
So the platform is built to do two jobs at once. It cuts response times and deflects routine questions to automation, which is the cost story every finance team wants to hear. And it tries to convert - nudging a billing question into an upgrade, a complaint into a save. The company frames it bluntly: turn cost centers into profit centers. Whether every deployment lives up to that is a separate question, but it explains why LivePerson sells to marketing and revenue teams, not just support desks.
03 / WHO USES ITThe invisible infrastructure behind big brands
LivePerson's customer list reads like a directory of institutions you interact with but rarely think about: banks, telecoms, retailers, airlines. Names attached to the platform over the years include HSBC, Orange, and The Home Depot. These are not companies experimenting with a chatbot on a landing page - they are running millions of conversations, in multiple languages, with compliance and uptime requirements that rule out most lightweight tools.
That is the shape of LivePerson's business: fewer, larger customers, deeply embedded. It is enterprise software in the truest sense - once a bank wires its customer conversations through your platform, ripping it out is a project no one volunteers for. The stickiness is the point.
04 / HOW IT'S DIFFERENTThe moat is the conversations it already had
There is no shortage of companies promising conversational AI in 2026. Genesys, NICE, Twilio, Zendesk, Salesforce, Intercom - the contact-center and messaging market is crowded, and the arrival of large language models flattened some of the old advantages. Anyone can wire an LLM to a chat window now.
LivePerson's answer is that the model is not the moat. The moat is the volume of real, messy, enterprise-grade conversations it has already handled - the raw material for understanding what customers actually ask and how to answer them at scale. In 2023 it layered generative AI and LLMs onto the Conversational Cloud under a "trustworthy AI" framework, the operative word being trustworthy: for a bank, an AI that confidently invents a wrong answer is worse than no AI at all. Selling safety and control to regulated industries is a narrower pitch than "we have a chatbot," and a harder one to copy.
05 / THE STORYThirty years, thirteen acquisitions, one proxy fight
LivePerson's history is a lesson in survival. It went public on the NASDAQ in April 2000 at $8 a share - one of the last technology IPOs before the dot-com market collapsed and took most of its class down with it. LivePerson did not go down. It kept buying: HumanClick in 2000, Kasamba in 2007, and a run of AI-focused deals in 2021 - e-bot7, VoiceBase, Tenfold - that bolted on chatbots, voice analytics, and telephony. Thirteen-plus acquisitions in all, each one a piece of the conversation the company wanted to own.
The reinventions were not always smooth. Live chat became persistent messaging around 2016 with a product called LiveEngage, later rebranded the Conversational Cloud. Leadership turned over - founder Robert LoCascio eventually stepped away, and in 2024 John Sabino, formerly VMware's chief customer officer, took over as CEO. In an unusual twist that year, LoCascio filed a proxy statement nominating directors to the board of the very company he had built and left. Founders, it turns out, do not always let go quietly.
06 / WHAT'S NEXTVoice meets messaging
In April 2026, the story took its sharpest turn yet. SoundHound AI - a company built on voice recognition and voice agents - agreed to acquire LivePerson. The stated logic is neat: SoundHound talks, LivePerson types, and customers do both. Combined, the pitch goes, they can offer one AI that meets a customer whether they speak into a phone or thumb a message, an end-to-end omnichannel platform where the old wall between voice and text disappears.
The financial terms were sobering for a company of LivePerson's history - an implied equity value of roughly $43 million, about $3.33 a share, with the deal expected to close in the second half of 2026. It is a modest number for a business that once carried a much larger valuation, and a reminder that surviving three decades and building real infrastructure does not guarantee the market rewards you for it. What LivePerson built, though, does not vanish in an acquisition. The billion monthly conversations keep flowing. They just get a new owner - and, if the thesis holds, a voice.