I first used LivePerson when I was at Network Solutions and the product was a small chat application we could test on a site. A visitor typed. An agent answered in the same window. That is still the work LivePerson sells, now as Conversational Cloud, across website chat, mobile, SMS, and social messages. The company has said that traffic runs at nearly a billion messages a month.

SoundHound announced today that it has completed the purchase. LivePerson common stock left Nasdaq the same day. John Collins, who ran finance and operations at LivePerson, is chief financial officer of the combined company. The close note puts 25 Fortune 100 names and more than 750 patents on the combined roster, and says SoundHound retired LivePerson’s notes so the balance sheet now carries no debt (SoundHound, 2026).

The April price bought accounts and a debt cleanup.

When the companies announced the agreement on April 21, SoundHound valued LivePerson equity at about $43 million, a 22 percent premium to the prior month of trading. Counting the discounted debt, the package sat near $250 million. SoundHound expected about $74 million of LivePerson cash at close, before the 2026 convertible notes. Management then said 2027 revenue should land at least at $350 million to $400 million, with $100 million or more from LivePerson accounts, and that the same installed base could support more than $500 million if SoundHound sells voice into those accounts and sells LivePerson’s messaging software into SoundHound accounts (SoundHound, 2026).

LivePerson stockholders approved the deal on September 2 after an earlier meeting lacked enough turnout (LivePerson, 2026). SoundHound’s own second-quarter revenue was $61.9 million, up 45 percent, with a raised 2026 outlook of $230 million to $260 million that still excluded LivePerson (SoundHound, 2026).

SoundHound already ran chat. LivePerson ran it for more brands at higher volume.

I have followed SoundHound through the same years. It started as a music identification app and a voice layer for cars. Then it bought restaurant phone and drive-thru voice, Amelia for enterprise service desks, and Allset for ordering. Amelia already let a customer switch between talking and texting in the same case. OASYS, launched in May, listed phones, text, and web chat as places an agent could run. LivePerson adds the volume, the social and mobile connectors large brands already use, and the customer-success teams that sit on that software.

SoundHound also sits in restaurants, cars, and hotels. Resorts World Las Vegas used Amelia to run a digital concierge named RED. The Amelia purchase in 2024 was the first large step into banks and healthcare service desks. I covered that shift when OASYS shipped in May and earlier when the company moved Amelia into cars and restaurants (SoundHound Solves the Hardest Part of Running AI Agents; Voice Economy; December 2025 car assistants).

I still treat the in-car assistant in my Toyota as a programmed list. It answers the phrases it was taught. It does not guess what I will ask next. SoundHound has been selling an assistant that understands a request, predicts the follow-up, and completes the task. LivePerson puts that company next to the website and SMS queues those same large brands already run.

Deflection counts a call that left the queue. It does not tell you the customer’s problem was solved.

OASYS lets a team build an agent and test it before a customer hears it.

The close announcement names OASYS, the Orchestrated Agent System SoundHound launched in May, as the product that will carry LivePerson. The public OASYS page says teams can build agents from prompts and past conversation records, then simulate thousands of conversations and review the transcripts before anything reaches a customer. Humans can run those tests. Software can run them too. Buyers should ask to see that step.

Plenty of teams still treat an interactive voice response tree as an agent. A phone tree routes a caller through buttons. An agent is supposed to understand the request and finish the work, or hand it to a person with the context attached. The test cases are how you find out which one you bought.

I would keep LivePerson chat where it already sits and SoundHound voice where it already sits, then pass the customer record, the transcript, and the next action between them. Waiting for one merged console is a long program. The useful check is whether a chat session can continue as a voice call, and whether a voice call can continue as SMS, without that wait.

Restaurant voice, car assistants, hotel concierge work, bank service desks, and enterprise messaging already share a roster. LivePerson adds customers who already run that messaging software. The work after close is whether those customers stay, and whether a case can move from chat to voice.

Compare this pair with the customer experience tools you already buy.

SoundHound can now offer voice to LivePerson accounts, the request those accounts made most often, and offer LivePerson’s messaging software to SoundHound accounts. That is the $500 million story in the close note. The operational questions are narrower. How many LivePerson logos renew in the first two quarters. Whether those accounts turn on voice. Whether OASYS is the console both products share.

About Shashi Bellamkonda

Loading bio…

LinkedIn Profile

Get new posts by email:

Subscribe

Powered by