Salesforce is betting that the next big customer service battleground is the phone call. On November 2, 2026, the company announced it had signed a definitive agreement to acquire Cresta, a voice-first customer experience platform whose AI agents resolve complex support calls end to end. The deal is expected to close in the first quarter of Salesforce’s fiscal 2028 — February to April 2027 — subject to regulatory approvals and customary closing conditions. Salesforce did not disclose the price, although The Information reported it at about $2.7 billion, a figure Salesforce has not confirmed.
What Salesforce announced
Salesforce described Cresta as a voice-first operating system for customer experience. The plan is to pair Cresta with Service Cloud, so that each call is tied to the customer record and to the outcome it produces — whether that is a fix, a saved account, or a sale. Kishan Chetan, executive vice president and general manager of Agentforce and Service Cloud, called voice the most difficult and consequential service channel. Cresta CEO Ping Wu said the combination would embed real-time conversation intelligence and agentic experiences directly into Salesforce’s customer-service platform.
What Cresta actually does
Cresta’s platform has three parts that work together. Its voice agents hold natural, real-time conversations, handle interruptions, and resolve complex calls without handing off. Its Agentic Assist coaches human representatives during live calls with hints, reminders, and workflows tailored to each conversation. And its Conversation Intelligence layer analyzes every interaction, tracks topics and outcomes, and auto-scores agent behavior across human and AI-led calls. The agents can also connect to CRM, billing, and scheduling systems to retrieve information and complete tasks, using shared memory that preserves context across channels and handoffs. This is part of the same shift seen as AI agents now start 40% of GitHub pull requests, with software moving from suggesting work to doing it.
Why voice is the hardest service channel
Chatbots had years to mature in text, where a pause is invisible and a mistake is easy to correct. Voice is unforgiving: a caller expects an instant, natural reply, gets frustrated by awkward pauses, and may be calling about a high-stakes problem such as a cancelled flight, a billing dispute, or an outage. That is why Salesforce framed this deal around the phone rather than another chat channel. Cresta’s agents already run at enterprise scale for brands including Marriott, Comcast, and Alaska Airlines, which gives Salesforce proven voice deployments rather than a laboratory demo.
What happens to existing Cresta customers
For buyers, one sentence in the announcement may matter more than the strategy talk. After closing, Cresta will continue supporting the contact-center and telephony platforms its customers already use. A company running its phones on another vendor’s software would not have to switch systems to keep the product. That commitment also matters for trust: just as AI agents automating security fixes still need human oversight and testing, voice agents handling live customers need guardrails, supervision, and a clean handover to a person when a call goes wrong. Cresta says its design combines reasoning with deterministic code that tracks conversation state, enforces required workflow steps, and lets staff guide a response or take over.
The price question: reported $2.7 billion, undisclosed officially
Treat the price with care. Salesforce disclosed no financial terms. The reported $2.7 billion figure comes from The Information, not from Salesforce, and would value Cresta at roughly 27 times its reported $100 million in annual recurring revenue. Cresta raised a $125 million Series D in November 2024, lifting its total funding above $270 million, and was founded at Stanford in 2017. Until Salesforce files deal documents or confirms terms, the $2.7 billion number is a report, not an official price — and readers should not quote it as confirmed.
What this means for businesses and callers
For businesses, the deal points to a future where the first voice on a support call may be an AI agent that already knows the customer’s history, can complete the task, and can hand over cleanly with full context. For callers, the test will be practical, not promotional: shorter waits, fewer repeats of the same story, and resolutions that actually stick. The risk is equally practical. Voice agents can misunderstand accents, miss emotion, or confidently give a wrong answer, and as AI phishing defenses remind us, voice is also a channel attackers exploit with cloned audio. Companies adopting voice AI will need to publish clear escalation paths, protect call recordings and transcripts, and measure resolution — not just containment — before expanding automation.
What to watch before the deal closes
Three things will decide whether this acquisition pays off. First, regulatory clearance and the closing timetable in early 2027. Second, how cleanly Cresta integrates with Service Cloud without disrupting the contact-center platforms its customers already run. Third, whether Salesforce keeps Cresta’s coaching and analytics as a neutral layer across vendors, or gradually steers customers toward its own stack. The deal also lands less than two months after Salesforce closed its purchase of Fin, formerly Intercom, on September 10, making Cresta the latest in a run of eight definitive agreements since June. Voice, it seems, is where Salesforce’s service ambitions now converge. Developers already work alongside AI coding assistants in 2026, so the question is less whether agents can handle work than how well they are supervised.
