Salesforce agrees to buy Cresta for voice-driven customer service AI – Unite.AI

Salesforce announced on November 2, 2026, that it has signed a definitive agreement to acquire Cresta, whose platform combines voice agents, human representative support, and conversation analytics. The transaction is expected to close in the first quarter of Salesforce’s fiscal 2028, subject to required regulatory approvals and customary closing conditions.

Salesforce described Cresta as a voice-first customer experience operating system, with Kishan Chetan, executive vice president and general manager of Agentforce and Service Cloud, calling voice the most difficult and consequential service channel. Cresta CEO Ping Wu said the combination will embed real-time conversational intelligence and agent experiences directly into Salesforce’s customer service platform.

Voice agents connected to company systems

Salesforce said Cresta’s voice agents resolve complex customer calls end-to-end, hold natural conversations in real time and handle interruptions. Marriott, Comcast and Alaska Airlines are among the brands using the platform on an enterprise scale. These capabilities are the operational element that Salesforce wants to connect to Service Cloud customer history and systems to get the job done, with resolution, retention, and revenue as the stated business outcomes.

Cresta’s AI agent documentation describes shared memory that preserves context across channels, transfers, and transitions between human agents and AI agents. Its agents can connect to CRM, billing, scheduling and other systems to retrieve information and complete tasks, using secure function calls and MCP support to access tools. In practical terms, the software is designed to both carry a conversation and run the service workflow behind it.

Deployment begins with existing conversation data. Cresta’s automation planning evaluates topics based on frequency, complexity, and resolution outcomes to identify candidates for AI agents, while successful interaction analytics is used to incorporate behaviors associated with resolution, customer satisfaction, and conversion. The approach links automation decisions to patterns in the customer’s service history, rather than treating each call type as equally suitable for an AI agent.

Real-time support for human representatives

The transaction also covers tools for people who run the live service. Cresta’s Agentic Assist provides real-time suggestions, reminders and workflows tailored to every conversation. Its insight system combines what is said with on-screen context to produce source-supported responses, while customizable AI summaries capture structured information and reduce post-call work. The product therefore supports a mixed operating model where AI can act independently or assist a representative during the same customer journey.

Cresta’s Conversation Intelligence extends this model to quality and training. It analyzes interactions driven by humans and AI, tracks customer topics and outcomes, and automatically evaluates agent behaviors in every conversation. Quality and outcome data fuel personalized coaching plans, while simulations based on real customer conversations give reps the ability to practice specific behaviors. The result is a shared layer of analytics for human and automated services, linking assessment and coaching with the improvement of AI agents.

Distribution controls and existing infrastructure

The Cresta controls are a material part of that package. The AI ​​agent design combines reasoning with deterministic code that tracks the state of the conversation and applies the required workflow steps. System-level guardrails, real-time supervision and monitoring models sit alongside an operations center where staff can lead a response or take charge of a conversation.

Before changes are promoted, agents can be tested against synthetic clients derived from conversational data, production errors can become test cases, and controlled A/B experiments can compare versions on real-time traffic. These mechanisms address the operational problem of modifying an automated service system without losing sight of the quality, compliance, or behavior experienced by customers.

Salesforce’s commitment post-closing is that Cresta will continue to support the contact centers and telephony platforms already used by its customers. This commitment defines the practical stakes of the acquisition: Salesforce is adding autonomous voice services, human assistance and interaction analytics to its customer services portfolio, while promising continuity for existing deployments. The central question of the integration is whether these functions can be brought together in Service Cloud without disrupting the infrastructure on which Cresta customers depend.