6 insights service leaders need to know about agentic AI (www.zdnet.com)

Salesforce’s seventh State of Service report, distilled through an interview with Michael Maoz, signals that agentic AI—systems that sense, reason, decide and act—is moving from experiment to production in customer service. Consumer-facing industries (finance, travel, retail) lead adoption, with Salesforce citing examples like 85% autonomous resolution on some service cases and the Agentic Enterprise Index forecasting ~50% of service cases resolved by AI by 2027. Leaders call AI-agent investment essential; 79% agree, 88% prioritize tech integration, and 44% report data silos have already delayed AI projects. The report also highlights an accelerating shift toward multimodal interactions (text, voice, images, sensor data) and broad use of generative, predictive, and agentic models to personalize and speed outcomes. Technically, the story is about action-oriented AI (Salesforce Agentforce) moving beyond content generation to execute workflows—refunds, rescheduling, returns, form fills—via prebuilt and custom agents tied to unified knowledge and CRM data. Key implications: clean, connected, discoverable data and strong knowledge management are prerequisites; agentic automation can cut service costs (~20%) and reduce average handling time while freeing reps from repetitive work so they can handle complex, high-value interactions (today reps spend ~46% of time on customer engagement). The takeaway for AI/ML teams: prioritize data integration, embed multimodal capabilities, instrument outcomes (AHT, resolution rate, CSAT), and design automation with human-centered governance and upskilling in mind.