Salesforce Signs Deal to Buy Cresta, Betting on Voice Agents for Service

Salesforce said Monday it has signed a definitive agreement to acquire Cresta, a startup whose voice agents resolve complex customer calls end to end and whose software coaches human representatives in real time. The company didn't put a price on the deal. It expects the transaction to close in the first quarter of fiscal 2028, subject to regulatory approvals and customary conditions — a window that falls between February and April 2027, because Salesforce's fiscal year starts Feb. 1.

The deal is Salesforce's second service-AI acquisition in under two months, and its eighth definitive agreement signed since the start of June. It signals where the company thinks customer service is heading: not just chatbots on websites, but software that can hold its own on a phone line.

What does Cresta actually do?

Cresta's voice agents run at enterprise scale for Marriott, Comcast and Alaska Airlines, according to Salesforce. They're built to hold real-time conversations and handle interruptions — the basic mechanics that trip up most automated phone systems. The same platform does three jobs at once: it coaches live representatives, it tests and tunes the voice agents, and it grades both humans and machines against one quality standard.

That unified scoring matters for buyers. Instead of running separate quality metrics for people and software, a contact center can compare them directly, and tune each against the other.

Salesforce plans to pair Cresta with Service Cloud, so that each call is tied to the customer record and to what came of it — whether that's a fix, a saved account or a sale. The pitch is continuity: the call doesn't disappear into a telephony log, it becomes part of the customer's history inside Salesforce.

"Voice is where service is hardest to get right, and where it matters most to customers," said Kishan Chetan, EVP and GM of Agentforce and Service Cloud at Salesforce.

Why voice, and why now?

Cresta lands less than two months after Salesforce closed its purchase of Fin, the company formerly called Intercom, on Sept. 10. Fin's customer agent works across live chat, email, WhatsApp, SMS, voice and Slack for more than 30,000 companies, and Salesforce said it resolves 76% of queries on average.

Cresta is built for the phone specifically. Salesforce called the phone the place customers turn to for their highest-stakes issues — the billing dispute, the canceled flight, the angry account on the edge of churning. Text-based agents handle volume; voice agents, if they work, handle consequences.

For buyers weighing the deal, one line in the announcement may count for more than the strategy talk. After closing, Cresta will keep supporting the contact center and telephony platforms its customers already use, Salesforce said. A company running its phones on another vendor's software wouldn't have to move to keep the product. That removes the biggest objection to any service-tool acquisition: rip-and-replace.

Cresta CEO Ping Wu framed the plan as embedding "real-time conversation intelligence and agentic experiences directly into Salesforce's first-class customer service Platform."

How big is the buying spree?

The Cresta agreement adds to a long run of shopping. Salesforce's own acquisitions page lists eight definitive agreements signed since the start of June:

• Contentful
• m3ter
• Fin (closed Sept. 10)
• Listen Labs
• Loops
• Braze
• Rune Technologies
• Cresta

The Braze agreement, signed Oct. 20, carried about $5.66 billion in enterprise value. Rune followed nine days later. Cresta makes it three service-related deals in a stretch of months, alongside a broader pattern of consolidation across marketing, communications and support software.

The buying spree comes with guardrails. Last week, Salesforce added cost and safety controls for AI agents, including budget caps and a kill switch in Agent Fabric — a sign that the company is thinking about runaway agent costs as it deploys more of them across its platforms.

Where did Cresta come from?

Cresta raised a $125 million Series D in November 2024. WiL (World Innovation Lab) and the Qatar Investment Authority led the round, with Andreessen Horowitz, Greylock Partners and Sequoia Capital among the returning investors. The round lifted its total funding above $270 million.

At the time of that raise, Cresta said it had nearly quadrupled annual recurring revenue over the previous two years. The startup's trajectory — enterprise customers like Marriott and Alaska Airlines, real-time coaching for human reps, then autonomous voice agents — tracks the industry's shift from assisted AI to agentic AI in the contact center.

Salesforce hasn't disclosed what it's paying, and it may not before closing. What's clear is the strategic logic: Service Cloud is Salesforce's largest cloud by revenue, and voice remains the channel where automated service fails most visibly. Buying Cresta, after buying Fin, gives Salesforce agents across both text and phone under one customer-record umbrella.

What happens next?

The deal needs regulatory approvals before it can close, with Salesforce guiding to that February–April 2027 window. Until then, Cresta operates as-is, and its customers stay on their existing telephony platforms regardless of vendor.

The open question is integration. Salesforce now owns a text-first resolution engine in Fin that averages 76% query resolution, a voice-first platform in Cresta graded against the same quality bar as human reps, and its own Agentforce stack — plus cost controls in Agent Fabric added just last week. If the company can tie all three to the customer record in Service Cloud, it can argue it owns the full arc of a service interaction. That's the bet buyers will be watching for when the deal closes in 2027.