Customer service is entering a new phase of enterprise transformation defined less by incremental automation and more by the rise of AI agents acting as digital labor. Salesforce’s 7th annual State of Service report, based on a global survey of 6,500 service professionals, shows how quickly this shift is unfolding.
By 2027, survey respondents expect AI to resolve half of all service cases, up from the 30% they estimate is handled today. That projection alone marks a fundamental change in how organizations expect work to be done across customer service, field service, and the broader enterprise.
“Service is where the pressure is highest and the margin for error is smallest,” Kishan Chetan, EVP and GM of Salesforce Service Cloud, tells ERP Today. “You’re dealing with real customers, in real time, across every channel. That environment is forcing organizations to rethink how work actually gets done.”
That pressure helps explain why service has become one of the most visible proving grounds for agentic AI. Service organizations face rising customer expectations, persistent cost pressure, workforce shortages, and fragmented systems simultaneously. In response, they are adopting AI faster and at greater depth than many other functions, setting patterns ERP leaders and enterprise architects will increasingly be asked to replicate elsewhere.
The report shows this shift is being driven largely by agentic AI capable of acting autonomously within defined workflows. These agents are no longer limited to answering FAQs. They are increasingly expected to manage order inquiries, summarize conversations, retrieve knowledge for representatives, and make personalized recommendations while maintaining context across channels.
AI Becomes a Strategic Priority
One of the clearest signals in the report is how rapidly AI has moved up the service leadership agenda. In just one year, AI jumped from the tenth-ranked priority to the second, trailing only improving the customer experience. That matters because customer experience itself has not changed as a goal. What has changed is how service leaders believe it can be delivered.
“AI is no longer a back-office efficiency tool,” Chetan says. “Agents can understand context, take action, make decisions, and adapt in real time. That changes the role of human representatives, giving them space to focus on complex, high-stakes interactions where judgment and trust really matter.”
This shift is already visible in operational expectations. Organizations using AI agents anticipate average reductions of 20% in service costs, case resolution times, and customer wait times, alongside gains in customer satisfaction and upsell revenue. In life sciences and biotech, expected upsell gains reach 20%, highlighting how service interactions are increasingly viewed as revenue-adjacent rather than cost centers.
Connected Systems Determine AI Success
Despite growing enthusiasm, the report is explicit about what separates successful AI deployments from stalled ones.
Data integration stands out as one of the strongest predictors of success, as 44% of service leaders say technology silos have delayed or limited their AI initiatives. Organizations that integrate service channel data into a single unified platform are 1.4 times more likely to describe their AI implementations as “very successful.”
For Chetan, this is where service begins to reshape enterprise architecture more broadly.
“For years, enterprises treated service as something that could sit on top of fragmented systems,” he says. “That no longer works. Service is becoming the forcing function for enterprise architecture transformation. When customer data lives in one place, service history in another, and product data somewhere else entirely, AI agents simply can’t deliver the contextual service customers expect.”
Service leaders, he adds, cannot afford to wait for multi-year modernization programs when customers expect immediate, personalized responses across channels. As a result, they are pushing for unified data platforms that connect customer, product, asset, and service histories in real time. Once those foundations are in place for service, other enterprise functions tend to follow.
The report reinforces this point quantitatively, as 88% of service leaders say they are prioritizing technology integration to support AI. That suggests that service organizations are increasingly shaping how enterprises think about data models, integration layers, and governance frameworks, particularly as AI agents begin to act autonomously.
Field Service Highlights ERP Intersection
Field service offers one of the clearest illustrations of how service-focused AI intersects with ERP systems. According to the State of Service report, field technicians spend 18% of their working hours on administrative tasks, more than seven hours per week that could otherwise be spent on repairs and maintenance. Scheduling conflicts, travel time, waiting for parts, and switching between disconnected applications all contribute to lost productivity.
“Every minute a technician spends on paperwork is a minute a customer is waiting,” Chetan says, noting that asset-intensive businesses feel this strain most acutely when service level agreements depend on fast, first-time fixes.
AI adoption in field service is accelerating in response, as 85% percent of field service leaders expect their AI investments to increase over the next year. Nearly all plan to use AI for instant access to information through knowledge retrieval, while many are exploring visual diagnosis and augmented reality-guided repairs. Technicians estimate that AI agents could handle 35% of their administrative work, saving up to 14 hours per week.
Those gains depend heavily on ERP integration. “Our field service capabilities are specifically designed to integrate with existing ERPs and other critical systems used by in-the-field workers,” Chetan explains. “An AI agent that identifies a failing part needs real-time inventory visibility. Predictive maintenance requires asset history and maintenance records. Intelligent scheduling requires workforce data from ERP.”
To support that orchestration, Salesforce has taken an API-first approach powered by MuleSoft. “We pull data from and push updates to the core systems that run the business,” Chetan says. “The real opportunity is creating closed-loop intelligence, where AI agents can identify patterns and continuously feed insights back into procurement, HR, and finance.” Services shift from reactive into “proactive, optimized operations across the enterprise,” with service agents both consuming and updating data continuously.
Conversational AI Raises Expectations
Another theme running through the report is the maturation of conversational AI—36% of organizations using both voice and text AI have enabled true multimodal interactions that preserve context across channels. Among service professionals using conversational AI:
- 89% say it increases self-service resolution
- 88% say it accelerates resolution times
- 85% say handoffs to human representatives are seamless.
These results matter beyond service because they reset expectations for how users interact with enterprise systems. Service environments provide some of the fastest and most demanding feedback loops for AI performance.
“Service is a great testbed for AI across the enterprise because it’s a high-frequency, high-stakes environment,” Chetan says. “It touches sales, marketing, product, operations, finance, and more. The ROI is immediate and measurable.”
Once organizations see AI working reliably in service, it quickly becomes a reference point for what they expect elsewhere in the enterprise. As Chetan puts it, “Service is the canary in the coal mine for enterprise AI.”
Service as Blueprint for Agentic Enterprise
Salesforce’s State of Service report positions service organizations as early architects of the agentic enterprise. They are confronting rising customer expectations, workforce constraints, and system fragmentation under intense pressure. In response, they are adopting AI not as a feature upgrade, but as an operating model change.
“The organizations getting this right aren’t treating AI as a technology project,” Chetan says. “They’re redesigning how decisions are made, how work is flowing across systems, and how humans and agents collaborate.”
By 2027, if half of service cases are indeed resolved by AI, the implications will extend well beyond contact centers. Data architectures, ERP integrations, workforce strategies, and governance models will all need to adapt to a world in which digital agents act alongside humans across core processes. Service is simply where those pressures are most visible today.
What This Means for ERP Insiders
Service is driving enterprise architecture change. AI-heavy service operations are pushing organizations to confront long-standing data and integration gaps sooner than other functions. As service teams push for real-time, contextual AI, they are accelerating demand for unified data models and platforms that can support autonomous decision-making across the enterprise.
Agentic service raises the bar for ERP integration and governance. As AI agents move from assisting representatives to resolving cases and orchestrating field service workflows, ERP systems responsible for assets, inventory, scheduling, and compliance become active participants in real-time operations. This shift increases the importance of API-driven integration, explainability, and auditability, as governance frameworks must operate at the same speed as autonomous processes.
Workforce impact is shifting from displacement to redesign. The report findings suggest that AI is reshaping service roles toward higher-complexity, judgment-driven work, influencing how enterprises think about skills development and career paths. ERP leaders should treat AI adoption as an operating model and workforce strategy challenge as much as a technology initiative.
