Salesforce to Acquire Cresta for Voice-Led CX AI

Salesforce has agreed to acquire Cresta, a company specializing in voice-led artificial intelligence for customer service operations. The deal brings Cresta's real-time conversational AI technology into Salesforce's growing portfolio of customer experience tools.

Cresta's technology is built to support live voice interactions, helping contact center agents and automated systems respond to customers in real time. Folding this capability into Salesforce signals a deeper push by the company into voice as a primary channel for AI-assisted service, alongside its existing text and digital-first offerings.

Voice remains one of the hardest channels to automate well — it's unscripted, emotionally loaded, and happens in real time with no room to edit a response before it lands. An acquisition like this suggests Salesforce is betting that voice-led AI is becoming table stakes for enterprise customer service platforms, not a niche add-on. For large organizations running call centers, it points to a near-term future where live voice interactions are coached, monitored, or partially automated by AI embedded directly in the CRM layer, rather than bolted on through a separate vendor.

For leaders building AI and CX strategy, the move is a signal to watch: platform consolidation in voice AI is accelerating, and the gap between "chat AI" and "voice AI" maturity is closing fast. Buying rather than building also shows how quickly specialized AI capability is being absorbed into general CX suites.

Deals like this get framed as a technology story, but the real test is always what happens on the call itself — not in the roadmap slide.

Voice is the channel where AI promises have historically outpaced delivery, because tone, hesitation and frustration are much harder to model than typed text. The operators who win here won't be the ones who deploy voice AI fastest — they'll be the ones who design for the moments it should stay quiet and let a human take over. Buying capability is easy; building the judgment for when to use it is the actual differentiator.