Agentic AI is redefining what it means to “serve” a customer. OpenAI’s Greg Brockman says the era of the chatbox is over – here’s what comes next, and what it means for every CX leader still betting on the chat interface.
OpenAI’s co-founder Greg Brockman has a simple message for anyone still debating whether AI chatbots will replace human customer service agents: you’re asking the wrong question. In a recent interview on the Big Technology Podcast, Brockman articulated a vision that goes well beyond smarter bots and faster response times. The shift, he argues, is architectural – from AI that talks to AI that acts. For the customer experience industry, that distinction is everything.
Agentic AI – autonomous systems capable of reasoning, planning, and executing multi-step tasks without human hand-holding is no longer a roadmap item. According to Gartner, it will automate 80 percent of customer service queries by 2028 while reducing costs by 30 percent. Cisco research puts 68 percent of all customer service interactions in the hands of agentic systems within the same timeframe. The chat widget on your website may already be obsolete — the industry just hasn’t caught up yet.
What Is Agentic AI – and Why Is It Different From Chatbots?
The distinction matters, and it’s sharper than most vendor marketing suggests. Traditional conversational AI operates within scripted boundaries. It matches intent to a predefined decision tree and, frustratingly, fails the moment a customer veers off-script.
Agentic AI operates differently. It understands goals rather than inputs, reasons through multi-step problems, uses tools (APIs, databases, calendar systems, payment processors) and executes actions without requiring a human to confirm each move.
As Zoom CX’s Head of AI Product Ram Rajagopalan put it:
“Instead of programming exact decision trees, you can set broad objectives and allow the AI to reason, adapt, and act autonomously based on real-time context.”
Brockman frames this transition as the natural next step for ChatGPT itself – a product he says has already outgrown its chatbox. OpenAI’s Codex, originally a developer-facing coding tool, is now embedded inside ChatGPT and used internally at OpenAI as routinely as Slack. The agent architecture it runs on – reasoning loops, tool use, persistent context – is no longer an engineering curiosity. It’s the product.
How Does Agentic AI Reshape the Customer Journey?
This is where the CX implications land hardest. The chat era was organized around interactions – discrete touchpoints measured in handle time, CSAT scores, and resolution rates. Agent-based AI operates on intent – what does this customer ultimately need, and what is the fastest, most friction-free path to getting them there?
That reframe cascades through every layer of the CX stack. Routing logic built around keywords and queues gives way to systems that understand context and act pre-emptively. Proactive service becomes viable at scale – an AI agent detecting a network outage, cross-referencing a CRM, and alerting affected customers before they’ve even thought to call, is no longer a theoretical exercise. Zoom reports automating 97 percent of its online customer queries using its own Virtual Agent – not through scripted flows, but through goal-driven reasoning.
For enterprise CX leaders, this isn’t a feature upgrade. It’s a workflow redesign problem.
Is the CCaaS Vendor Landscape Ready for Agentic AI?
The pressure on established platforms is real and growing. If an AI agent can span channels, query back-end systems, and resolve customer issues autonomously, what is the platform layer actually for?
The smarter vendors are getting ahead of this. Salesforce launched Agentforce Contact Center in March 2026, positioning itself as the first solution to natively unify voice, digital channels, CRM data, and agentic AI in a single architecture – a direct play for the intent-driven service model Brockman describes. Salesforce’s Agentforce 2dx goes further still, embedding proactive agentic triggers across enterprise workflows, not just customer-facing ones.
NICE and Genesys are navigating similar terrain, building agentic layers atop existing CCaaS infrastructure. But the architectural advantage belongs to whoever can most convincingly claim the “unified intent layer” – the system that knows what the customer needs, has access to the data to act on it, and doesn’t require a human to bridge the gaps.
OpenAI, through its Agents SDK and growing enterprise partnerships, is increasingly positioning itself as that layer. The WSJ reported that OpenAI is actively courting businesses to build proprietary agents on its platform. For legacy CCaaS vendors, that is a different kind of competitor than they’ve faced before.
What’s the Biggest Challenge Holding Agentic CX Back?
Brockman named it directly: trust. Specifically, the incremental, earned kind. How much authority do you grant an AI agent before a human must sign off? The answer varies by context – and in customer service, the stakes of getting it wrong are immediately visible.
Refund processing, complaint handling, high-value account queries, sensitive healthcare or financial interactions – these are not equivalent use cases. A one-size-fits-all autonomy threshold is not a strategy; it’s a liability. Gartner’s Daniel O’Sullivan describes the opportunity precisely:
“Agentic AI has emerged as a game-changer for customer service, paving the way for autonomous and low-effort customer experiences”.
The practical architecture here is a tiered autonomy model: define which interaction types the agent can fully resolve, which require a human review before acting, and which escalate immediately. That framework doesn’t exist out of the box in most platforms today, and building it requires collaboration between CX leadership, legal, and AI product teams that many enterprises are only beginning to establish.
What Should CX Leaders Do Right Now?
The worst response to Brockman’s vision is to wait and see. The second worst is to retrofit agentic AI onto workflows designed for interaction-based service. Rajagopalan’s advice cuts through the noise:
“Don’t start with the question, ‘Where can I use agentic AI?’ Start with, ‘Where are the pain points in my customer experience?'”
In practice, that means auditing your highest-volume, highest-friction interaction types – the ones your agents handle dozens of times a day – and asking whether the entire resolution path, not just the opening exchange, could be executed by an agent. It also means pressure-testing your vendor stack. If your CCaaS platform can’t articulate a credible agentic roadmap, or if its answer requires five integrations and a systems integrator, that’s a signal worth taking seriously.
The chat interface isn’t being improved. It’s being replaced. Whether CX teams lead that transition or are led by it is, at this point, a choice.
Frequently Asked Questions (FAQs)
What is agentic AI in customer service?
Agentic AI refers to autonomous systems that can understand a customer's goal, reason through the steps required to achieve it, and take actions such as processing a refund or updating an account without requiring a human to manage each step.
How is agentic AI different from a chatbot?
Chatbots follow scripted decision trees; agentic AI uses real-time reasoning and tool access to pursue goals autonomously, adapting to unexpected inputs rather than failing on them.
When will agentic AI dominate customer service?
Gartner forecasts agentic AI will handle 80 percent of customer service queries by 2028; Cisco puts the figure at 68 percent of vendor-facing CX interactions within the same window.
Is ChatGPT being used in enterprise customer service?
Yes. OpenAI's Agents SDK enables enterprises to build custom agents on its models, and the platform is actively positioning ChatGPT and Codex for business process and service automation.
What are the risks of deploying agentic AI in CX?
The primary risks are trust and liability: AI agents acting without appropriate oversight can mishandle sensitive interactions, creating reputational and legal exposure; tiered autonomy models and human-in-the-loop escalation paths are essential safeguards.
What CCaaS vendors are leading on agentic AI?
Salesforce (Agentforce Contact Center), NICE, and Genesys are among the established CCaaS players actively building agentic capabilities; OpenAI itself is emerging as a platform-layer competitor.
