The Agent at the Counter: Why AI's Next Job Is Customer Support, Commerce, and Coordination
AI agents are moving from conversational helpers to customer-facing operators that can coordinate support, commerce, and back-office workflows across enterprise systems.
The old customer-service chatbot had a narrow job: greet the customer, search a knowledge base, and hand off anything complicated to a human. The new AI agent is being designed for something more consequential. It is supposed to recognize intent, check account context, coordinate with other software, take action, and stay inside the rules while doing it.
That is why the latest enterprise AI announcements are less about conversational polish and more about plumbing. Google is pushing Agent2Agent, or A2A, as a protocol for agents to communicate across systems. Salesforce and Google Cloud are expanding integrations that let agents operate across Slack, Google Workspace, Agentforce, and Gemini Enterprise. Workday is putting governance around the same shift with Agent Passport, while NiCE keeps turning customer experience into an agentic automation problem.
From Chatbots To Counterparties
The important thing is not that an AI agent can answer a question about a return policy. The important thing is that the agent may be able to see the order, verify the customer, start the return, trigger a replacement, update the CRM record, and summarize the exchange for a human service team.
In commerce, that turns the agent from a widget into a counterparty. In customer support, it turns the agent from a script reader into an operational teammate.
Why Support Becomes The Testbed
This is the clearest reason companies keep returning to customer support as the testbed for agentic AI. Support is messy, repetitive, measurable, and deeply connected to revenue.
A failed interaction is obvious. A successful one can reduce cost, preserve a customer relationship, and generate data about what the product or buying experience is failing to explain.
Commerce Becomes Machine-Mediated
Agentic commerce changes the customer relationship too. Once an assistant can compare products, negotiate preferences, initiate a cart, and coordinate payment, retailers no longer only design for human browsing. They also design for machine-mediated buying.
A customer may not visit a product page at all. Their agent may ask another agent whether an item is in stock, whether shipping will arrive by Friday, whether a coupon applies, and whether the merchant's return policy satisfies a personal preference profile.
Governance Becomes The Product
That makes interoperability a business issue, not only a developer issue. Google's A2A announcement frames the problem plainly: agents need a way to coordinate across enterprise systems instead of remaining trapped inside single-vendor silos.
The useful middle ground requires identity, permissions, audit trails, tool boundaries, escalation rules, and human override. A customer-facing agent that cannot act is just a chatbot with better manners. A customer-facing agent that can act without limits is a risk.
What Workers Actually Change
For workers, this creates a more complicated picture than the usual automation story. The first visible impact may be less about replacing entire jobs and more about changing the shape of participation.
A support representative might move from answering every routine ticket to supervising exceptions, coaching agent behavior, handling emotionally sensitive cases, and tuning playbooks. A commerce manager might spend less time configuring static journeys and more time deciding what agents are allowed to do for which customer segments.
Sources
Google Cloud blog: Agent2Agent protocol upgrade, July 31, 2025: https://cloud.google.com/blog/products/ai-machine-learning/agent2agent-protocol-is-getting-an-upgrade
Salesforce press release: Salesforce and Google Cloud launch new integrations, April 22, 2026: https://www.salesforce.com/news/press-releases/2026/04/22/salesforce-google-cloud-launch-new-integrations-deep-context/
NiCE press release: Agentic AI innovation for customer experience, March 10, 2026: https://www.nice.com/press-releases/nice-launches-agentic-ai-innovation-that-turns-enterprise-interaction-data-into-ready-to-deploy-ai-agents
Workday press release: Agent Passport, June 2, 2026: https://investor.workday.com/news-and-events/press-releases/news-details/2026/Workday-Launches-Agent-Passport-to-Test-Verify-and-Continuously-Monitor-Every-AI-Agent-in-the-Enterprise/default.aspx