Salesforce Inc. today introduced its first fully autonomous AI agent for customer service.
Einstein Service Agent, built on the Einstein 1 Platform, lets customer service teams offload time-consuming tasks that bog down their productivity and compromise customer satisfaction.
Unlike traditional chatbots — which can only handle specific queries that have been explicitly programmed into their system and don’t understand context or nuance — Salesforce’s new agent can address a broad range of service issues without preprogrammed
scenarios, the company said.
Einstein Service Agent “will augment human co-workers and completely transform how service teams operate, making them far more efficient and productive,” Ryan Nichols, chief product officer, service cloud, at Salesforce, said in a video interview. “We are reimagining customer service for the AI era.”
Einstein Service Agent’s reasoning engine interacts with large language models (LLMs) by analyzing the full context of the customer’s message and then autonomously determining the next actions to take. It then uses generative AI to create conversational responses that are tailored to a company’s brand voice, tone and guidelines, according to Nichols.
For Salesforce, which has devoted a decade to predictive AI, the fully-autonomous AI agent represents a quantum leap in an age of trust in AI. “We are showing generative content to customers as part of a trusted AI approach,” Nichols said.
Einstein Service Agent is designed to respond to customers across self-service portals and messaging channels and perform tasks proactively while operating within clear guardrails, according to the company. When more complicated issues arise requiring escalation to a human co-worker based on the parameters set by the company, Einstein Service Agent hands off the task seamlessly.
“There is no frustrating dead-end, as is often the case with chatbots,” Nichols said. “You do have the option of human interaction.”
Einstein Service Agent will be generally available later this year. Though most companies use chatbots, 81% of customers would rather wait to speak to a live agent than chat with a bot because automated systems aren’t meeting their expectations and can be frustrating, according to an internal Salesforce study.
Still, 61% of customers say they’d rather use self-service to resolve simple issues, underscoring the need to deliver more capable automated services, Nichols said.
Customer service has been the most fertile area of AI adoption across Salesforce’s cloud operations because many customers already “have a rich and factual data set and knowledge articles within Salesforce to work with,” Rebecca Wettemann, CEO and principal analyst at Valoir, said in an email message.
However, the bigger question is whether Salesforce customers are ready, she cautioned.
In Valoir’s survey on AI and language this spring, for example, only 6% of workers believe AI should “automatically act for them without a human in the loop,” she said.
“Salesforce has done a lot to push the trust message, so it’s significant that it’s pushing the autonomous agent so quickly,” Wettemann wrote.
