What Salesforce said
The essay, titled Personal Agents Will Drive More Contact Center Traffic Than They Offset, appeared in Salesforce's newsroom. Its central idea fits in one sentence from Chetan himself: personal agents can start a customer's request, but the company still has to finish it.
Personal agents are AI assistants that a consumer hands tasks to: checking on an order, requesting a refund, rescheduling an appointment or comparing offers. According to Chetan, these agents:
- Don't get tired of waiting on hold: the hold time that wears out a person's patience doesn't affect them.
- Keep trying: if asked to track an order, they can call back every few hours until it ships.
- Contact several providers at once and at any hour, even in the middle of the night.
Multiplied by every consumer who delegates a task to an agent, Chetan writes, the volume can become enormous, and demand peaks could become the new normal. He also points out that voice is the most expensive service channel.
Salesforce's recipe and what remains to be proven
Chetan's proposal is that companies meet personal agents “at the door” with their own AI agents, because these are the ones that know the business's processes and data. According to Chetan, the personal agent lacks the customer's full context, runs into bot defenses and questions only a human can answer, and improvises, often without finishing the job. The second half of the text describes Salesforce products such as Service Cloud, Agentforce Contact Center and its assistant Casey.
PPC Land, which analyzed the essay on the same October 19, points out its limits: it includes no surveys, call volume figures, number of personal agent users or a date for the supposed increase, and its links lead only to Salesforce's own pages. It also notes that the headline is categorical while the body speaks in the conditional. It is, in other words, a reasoned prediction from a vendor that sells the solution, not a study.
The text follows another Salesforce essay from October 9, by Mark Wakelin, which warned companies that their customers' personal agent “is already knocking at the door”.
The context: agents are already calling and buying
The discussion is not theoretical. PPC Land cites several recent facts showing AI agents interacting directly with companies: on September 20, 2026, Amazon blocked Meta's Muse agent from placing orders in its store; security firm HUMAN estimated that Muse accounted for about 72% of the daily AI agent requests it observed; and AI voice platform Vapi said in May that it had handled more than one billion calls.
At the same time, agents are still imperfect. A 2025 Salesforce AI Research test, cited by PPC Land, found that agents completed 58% of simple business tasks and only 35% of those requiring several conversational turns. An agent that half-fails usually ends up in what Chetan describes: a new contact with the company.
What it means for businesses in Colombia and Latin America
Personal AI assistants are still emerging in the region, but it pays to prepare before the mass calls arrive:
- Size for peaks. If part of your contacts come from machines that keep trying, your ability to receive many calls at the same time becomes critical. Numbers that accept hundreds of simultaneous calls and well-configured queues keep your human customers from being left out.
- Automate the repetitive. Questions like “where is my order?” can be resolved by your own AI agent, by voice or on WhatsApp, at any hour.
- Share status before they ask. Proactive SMS or WhatsApp notifications reduce repeat inquiries, whether they come from people or agents.
- Set rules for dealing with bots. Decide which transactions an agent can carry out on a customer's behalf and how identity is verified, to protect your company from fraud.
- Measure. Check your reports for increases in repeat or after-hours calls; they will be the first sign.
At 19 80 TIC Cloud we combine high-volume virtual mobile numbers, a cloud call center and AI voice and chat agents so your company can handle peaks without losing customers.
