No Jitter is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

Salesforce Releases Next-Gen Service Cloud Einstein

Salesforce today announced the next generation of its Service Cloud Einstein, which combines its flagship CRM solution with artificial intelligence (AI) and guided workflows for agents and customers.

Specifically, Salesforce introduced three elements to Service Cloud Einstein:

  • Einstein Bots for Service -- chatbots incorporating machine learning and natural language processing for automating routine customer service requests and gathering customer information for agents; generally available, included in Service Cloud Unlimited Edition and Live Agent licenses
  • Einstein Next Best Action -- on the agent side, Einstein leverages rules-based and predictive models to deliver recommendations and answers to assist agents during customer interactions on any channel; on the customer side, Einstein will surface offers to customers at the right time in an interaction; in pilot, pricing information to come with general availability
  • Lightning Flow for Service -- guided service workflows agents can follow to remove some of the guesswork from customer service (think if customer says X, do Y), or that customers can use for help navigating through self-service processes; generally available, included with all Service Cloud editions

Bringing Intelligence and Processes Together

Delivering an engaging, digital experience requires an understanding of what customers want and how best to help them -- and not only on the business side but also from within an empowered contact center, said Bobby Amezaga, senior director of Salesforce Service Cloud product marketing, in a recent briefing. Delivering on these experiences requires a combination of intelligence and processes, which is what Salesforce is aiming to nail with this release, he added.

Because these new features and functionality are built on the Salesforce platform, they can be extended to partners to create packaged experiences for customers. In fact, the more data pulled in from third-party systems, the better the Einstein intelligent recommendations and guided processes will be, Amezaga explained. And it's when all of these features come together that you can see the true power, he added.

"Today, customer service really is no longer seen as a cost center," Amezaga said. "It's become a revenue driver, a competitive differentiator, and a reputation enhancer."

AI & Humans: Striking the Right Balance

Salesforce does a lot of user research, as Marco Casalaina, VP of product management for Einstein on the Service Cloud Einstein, told me during a demo of these new features. And it uses its findings to coach enterprises on how best to leverage AI in customer service scenarios. For example, the company has learned that organizations shouldn't try to represent bots as anything but bots. So it now recommends to customers not they not try to pass off a machine as a human, because consumers react negatively to that, Casalaina said.

As another example, Salesforce learned that a response from a chatbot shouldn't be immediate because users tend to "freak out," Casalaina said. For that reason, Salesforce has introduced a delay optimizer for bot responses and recommends organizations implement a few seconds delay.

In the demo Casalaina walked me through, an online banking customer opens a new credit card via a chatbot interaction. The chatbot then asks the customer to rate the conversation, and Einstein Next Best Action uses that interaction context and customer data to offer up action items like add spouse to the card, upgrade to platinum, or go paperless. In the demo, the customer clicked on "add spouse to card," at which point the bot escalated the transaction to a live agent.


On the bank's side, the agent who picks up the conversation sees the chat conversation between the customer and the bot. The intelligence built into the system provides the agent with a customized action list that includes the option to add a spouse to the card, as shown above in the left column.

Although not shown here, an Einstein Next Best Action prompt lets the agent know to follow the "upgrade to platinum" workflow after completing the add spouse request. Further, the agent can open a workflow chart showing which steps to take at various points in the interaction.

This demo served to illustrate another one of Salesforce's user research findings: The hybrid agent-bot model tends to work best for providing customer satisfaction in customer-agent interactions.

Related content:

Follow Michelle Burbick and No Jitter on Twitter!