Customer satisfaction outranked service rep productivity and average handle time in the list of key performance indicators (KPIs) most improved by agentic artificial intelligence, according to research published by Salesforce.
The San Francisco-based software provider surveyed more than 3,000 customer service professionals around the world to produce its report, State of Service: AI Agents Edition.
Salesforce found customer service departments’ use of agentic AI has risen by 39 per cent since last year, and is now common across 66 per cent of organizations.
Though AI agents can work autonomously in executing tasks, customer service professionals say the technology is changing how they work internally. More than half (52 per cent) say they are already rethinking existing processes, while another 45 per cent are planning to do so.
Service reps aren’t just waiting for the impact of AI. Salesforce noted the many ways they’re trying to prepare:
Much like similar studies, Salesforce’s report highlights the challenge of data readiness in deploying agentic AI. However there was a noticeable gap between the 59 per cent of leaders who said data readiness could be a barrier vs. the 70 per cent of reps who did the same.
Overall, 89 per cent of service professionals whose organizations are using AI say they see benefits in expanding deployments.
“AI agents are no longer just experiments. They’re actively embedded in workflows, handling routine tasks, surfacing insights, and freeing up human service reps to focus on higher-value work,” the report’s authors wrote. “Organizations are learning what works, where AI delivers the most impact, and how to build trust while scaling responsibly.”
360 Magazine Insight
Full disclosure: I provided content marketing services to Salesforce in Canada for more than 10 years and continue to do so in the U.S. However this was the first time I had seen this research. I was not asked to report on it and will not get any sort of payment for doing so.
It is in Salesforce’s interest to present AI-powered service in a positive light, but the report touches on potential risks and pitfalls too, including security issues and customer trust.
On the latter, Salesforce found 77 per centof companies with customer-facing AI allow customers to request a human representative at any point during an AI interaction. Though the report doesn’t say it, I will: this should be 100 per cent, and it should be the default standard for organizations across every single industry.
Some of the biggest value in the report goes beyond validating the technology and getting into the specifics of how it could change the way people work in customer service. For example, 66 per cent said they expected to see growth in data management roles. However 62 per cent simply forecast the rise of “specialist” roles, and 48 per cent cited “generalist” roles.
This suggests many organizations don’t fully know what kind of talent they’ll need to support customers alongside AI agents. The process changes cited earlier are telling: those changes should really be mapped out in advance of an AI deployment, or at least amid a pilot project.
Salesforce’s report is ungated and browser-based, and offers the ability to filter results by local geography to help customer service leaders create a more granular benchmark of how their peers in other organizations are using agentic AI.