Salesforce says “personal agents” could boost contact center call volume, pushing firms toward agent-assisted service workflows

Salesforce is warning customer service organizations that the rise of “personal agents” for consumers could create a new kind of workload for contact centers, even as automation improves. The company’s core point is straightforward: personal agents can begin a request, but businesses still have to complete it, sometimes across multiple steps, checks, and handoffs to humans.

The company describes how consumer-facing AI agents are being used for tasks that do not necessarily require a phone call, such as navigating a website, using application programming interfaces to perform actions, or reasoning through straightforward decisions. Examples cited include canceling subscriptions and booking reservations, where an agent might attempt to drive the process end-to-end without a human on the line.

However, Salesforce argues those capabilities break down when the task depends on information the personal agent cannot fully access. Even if an agent has access to an inbox and data a user explicitly shared, it may still lack the broader relationship context a business maintains, such as eligibility rules, loyalty tiers, or internal workflows needed to execute changes. The company also points to friction from security controls designed to block bot behavior, plus questions that only a human representative can answer.

Salesforce further says the contact center impact could be amplified by timing and volume. It notes that personal agents may accept waiting for hold times that frustrate humans, and they may repeatedly check on outcomes such as order updates or confirmation messages. Unlike people who call during business hours, AI agents can place calls at any time, which could make peaks in voice traffic more persistent.

In Salesforce’s framing, contact centers are already accustomed to spikes during peak seasons or other disruptions. The new risk is that agent-driven requests could shift those spikes into a steadier “new status quo,” particularly because voice remains among the costliest engagement channels. Salesforce says organizations should be prepared for a surge in “agentic traffic” even if the underlying customer behavior shifts away from direct dialing.

Salesforce’s recommended approach is to meet personal agents with business agents at the start of the interaction. The company argues that if a personal agent understands the customer’s request but lacks the operational tools and context to complete it, the interaction may stall or fail. A business agent, by contrast, is tied to defined processes and workflows and is grounded in the company’s actual customer data and system records.

The company’s services pitch centers on its Service Cloud platform and its agent tools for contact centers. Salesforce says Service Cloud was built to bring AI agents, human expertise, and trusted data together on one platform, so the right agent or representative can start from the customer’s complete record and case history. The company also points to its multi-channel coverage, saying its help agent, named Casey, can work across voice, web chat, SMS, WhatsApp, and Apple Messages so personal agents can continue the conversation regardless of channel choice.

Salesforce also outlines specific mechanisms it says support controlled automation. It describes Agent Script as a way to define what an agent can and cannot do, and it references a Testing Center for simulated test calls before an agent handles live conversations. When escalation is needed, Salesforce says Casey hands off to a human representative with an AI-generated summary, full transcript, and a recommended next action. It also cites Service Rep Assistant, which can generate step-by-step action plans for a case using real-time customer data and company knowledge, positioning human staff to work alongside AI rather than restart the work.

The company does not provide quantitative estimates of how much personal agents could increase contact center volume, nor does it disclose customer-level metrics or pilot results. It also does not specify which industries or geographies face the greatest voice impact. What is clear from its message is a strategic direction: treat agent adoption as a service-process problem, not only a chatbot problem, and invest in workflows that can absorb frequent escalations and asynchronous follow-ups.