A customer sends a question to a company’s help desk and gets an answer within seconds, with no human ever opening the thread. That is the outcome Fin sells: the customer-service AI agent says it resolves 76 percent of support tickets from start to finish without a person stepping in. On September 10, Salesforce paid roughly $3.6 billion to own it.
Salesforce said it had completed the acquisition of Fin, the agent formerly known as Intercom, and would fold it into Salesforce AI Labs to reinforce Agentforce, the company’s own push into AI-driven customer service. Fin carried more than 30,000 customers into the deal, a base that spans the same kinds of businesses Salesforce has spent two decades selling software to.
The purchase is not the only one in motion. Business Insider reported on September 9 that Salesforce is also in talks to buy Listen Labs, an AI platform for user research, for about $2 billion. The report landed a day before the Fin deal closed, a sign of a company moving on several fronts at once.
Listen Labs had already signed a $125 million Series C term sheet led by Menlo Ventures, at a valuation of $1.5 billion. The Salesforce talks tore up that signed document before the round could close. The startup runs at roughly $30 million in annualized revenue and counts Microsoft, Canva and Anthropic among its customers, according to the report.
At $2 billion, the Listen Labs price would work out to about 67 times annualized revenue. Whether that multiple survives Salesforce’s internal review is an open question, and it is the figure that will determine whether the talks turn into a signed deal.
Together, the two moves show how Salesforce is spending to keep its seat at the AI table. The company built Agentforce as its answer to the wave of generative AI that threatened to make traditional customer-relationship software feel dated, and it has been buying pieces to fill the gaps rather than building everything itself.
Marc Benioff, Salesforce’s chief executive, has cast AI agents as the successor to the chatbot era, arguing that businesses will soon run much of their customer service on software that acts rather than answers. The Fin acquisition hands that argument a working product with a large installed base and a hard metric, the 76 percent resolution rate, that Salesforce can show to skeptical buyers.
The customer-service agent market has become one of the most crowded corners of enterprise software. Startups such as Sierra and Decagon have raised billions of dollars, and every large cloud vendor has fielded a competing product. Consolidation is now beginning, with the biggest platforms buying the startups that proved the market could work.
Salesforce has a long history of such purchases. It built much of its growth on acquisitions that brought in products it then pushed through a sales organization few rivals can match. Fin gives that machine a new product to sell, and the 30,000 customers come attached.
The risk is the same one that has trailed every Salesforce acquisition: price. The company has been criticized before for paying rich multiples for growth that later slowed. A $3.6 billion price for Fin, and a possible $2 billion for Listen Labs at 67 times revenue, suggest the company is again willing to stretch.
Analysts said the deals matter less for the individual products than for the data and customers they bring. Every support ticket Fin resolves teaches the model something, and every Listen Labs account is a window into how companies test and refine products. In an AI race where training data and distribution count as much as code, that is what Salesforce is paying for.
The company has also been racing to keep enterprise customers from defecting to newer AI tools. Anthropic, which OpenAI and Salesforce alike are watching closely, has been courting large corporate buyers directly, and a coming public listing is expected to give it more capital to compete. Salesforce’s answer has been to bolt AI agents onto the customer data it already holds.
Fin’s strength is that it plugs into a company’s existing support stack and knowledge base, so the agent can answer from real records rather than generic text. That is the pitch Salesforce will make to its own customers, many of whom already pay for Service Cloud, the product the new agent would sit alongside.
Listen Labs, if the deal closes, would add a different capability: a way for companies to run AI-driven user research, the kind of testing that tells a product team whether a feature is worth shipping. That fits Salesforce’s broader effort to sell tools for the entire customer journey, from first contact to renewed contract.
Whether the Listen Labs talks end in a purchase remains unclear. The company could still walk away, the price could shift, or another bidder could surface. What is settled is the Fin deal, which closes one acquisition and signals that Salesforce intends to keep buying as the customer-service market is rebuilt around AI.
For now, Salesforce has added a product that resolves three out of four support tickets on its own and paid a price that reflects how much it wants to keep its customers from leaving for the startups that built those agents first.
