“Rebook my flight and get me home tonight. Keep me in the same cabin, avoid a connection and ask me before spending more than $500.” 

A traveler will soon be able to delegate that task to a personal AI agent and move on with their day. Their AI can liaise directly with the airline, evaluate alternatives and reengage the traveler when a decision requires their approval. 

Consumer AI agents such as Muse, Dots, Instinct and others are moving personal AI closer to widespread adoption. Organizations are already seeing AI arrive through channels curated for people including voice, web chat and websites. As these agents become more capable, brands will find themselves serving AI acting on their customer’s behalf. 

This has major implications for customer experience. Most journeys, systems and service models were designed for direct interaction between a person and a business. Personal AI introduces a new participant in AI capable of representing the customer, carrying their intent and taking action on their behalf.  

This raises questions enterprises haven’t historically needed to answer. Who is this AI agent? Who does it represent? What has the customer authorized it to do? And where does that authority end? 

My advice is to start preparing now.  

At Genesys, we believe organizations need to focus on three areas to manage personal AI at scale: interoperability, orchestration and governance.  

The Genesys Cloud™ platform is evolving for experiences that may begin with the customer or with an AI acting for them. We’re also working with Sierra, Meta and other industry leaders to develop the Personal Agent Protocol (Poppy), an emerging open protocol designed to create a trusted front door for personal AI.  

Poppy will help enterprises recognize a personal AI agent, authenticate the customer it represents and understand the authority that customer has delegated to it.  

Give personal AI a better way in 

Websites, apps, forms and contact centers were built for people. Personal AI can navigate them today, though they’re a clumsy interface for an AI agent. More importantly, they weren’t made to establish who an agent represents, what the customer has authorized it do or the limits of its authority.  

Personal AI shouldn’t have to pretend it’s a human to reach a business. Poppy creates a direct, structured way for personal AI to arrive as AI, establish who it represents and communicate what it has permission to do. 

That authority doesn’t have to be all or nothing. The customer controls what authority they delegate to their AI, and the enterprise determines what it makes available and under what conditions. Poppy provides the framework for communicating those permissions between them. 

Genesys is testing how this could work through proof-of-concept scenarios that progressively move from discovery to authorized action. In one scenario, a customer could ask Muse to move a flight.  

Muse will arrive through Poppy, Genesys will verify the agent and customer and the customer grants access to the booking. Genesys then connects the request to the airline’s API or workflow with identity, intent, permissions and relevant context staying attached until the authorized change is complete. 

This work is helping us test what the trusted enterprise entry point needs to support as Poppy develops its first version. 

I expect other protocols and approaches to emerge as this market develops. That’s a good thing. Businesses and consumers will benefit from an open ecosystem where personal AI can work across providers without requiring a proprietary connection between every agent and every enterprise. 

For brands, that means serving the customer’s chosen AI rather than requiring the customer to choose an AI based on which businesses it can reach. 

Prepare for AI on both sides 

Once personal AI has a trusted way into the enterprise, it requires an equivalent on the other side. 

Personal AI agents are persistent. They can research options, negotiate and return until they complete the task. Routing every exchange to an employee would miss one of the biggest opportunities AI creates. 

We need to let AI work with AI where it makes sense.  

As personal AI drives greater demand and interaction volume, enterprise AI can scale to manage that load by coordinating across systems and agents to drive customer outcomes. That’s why organizations need agentic orchestration to create a unified path for personal AI and enterprise AI to work together toward the customer’s goal. 

Poppy opens the interaction while Genesys Cloud provides the orchestration layer for the enterprise to fulfill the inquiry. 

When a request reaches the enterprise, the first job is to understand what the customer— or the AI representing them—is trying to accomplish. Genesys Cloud Navigator* supplies that intelligence by interpreting intent and using relevant customer and enterprise context to determine how the request should be handled. 

Genesys Cloud Contextual Intelligence brings shared enterprise memory to connect identity, interaction history, business events and real-time signals so decisions are grounded in an up-to-date understanding of the customer. 

Genesys Cloud Orchestrator* can then coordinate what happens next by selecting the appropriate AI agent, workflow, system, API or person and adapting the journey as circumstances change. 

For requests that fall within its scope, Genesys Cloud Agentic Virtual Agent can act as the enterprise AI counterpart and autonomously work with the customer’s personal AI toward resolution. 

Some interactions will happen entirely AI-to-AI. Others will require the enterprise to dynamically determine whether AI, an employee or another resource should handle each step. That flexibility matters because the right avenue will depend on the request.  

This creates a more deliberate role for people. Employees can focus on situations that call for judgment or additional authority leaving routine AI-to-AI interactions to the systems equipped to handle them efficiently. 

More AI requires stronger governance 

An agent acting for a customer needs explicit authority. 

Enterprise AI operates within its own set of policies and permissions. As work moves among personal AI, enterprise AI, systems and people both sides should understand what they’re permitted to do. 

Return briefly to our traveler. Their personal AI has permission to spend $500. If the best available flight costs $700, that authority is capped at the spending limit. The additional expense should be returned to the traveler for approval. 

Autonomy only works when everyone involved knows where that authority ends, and that’s why it cannot be separated from orchestration. 

The Genesys Cloud AI Control Plane provides the governance layer for autonomous environments. It combines AI discovery, identity, policy and observability together so organizations can maintain oversight as AI assumes consequential work. 

Personal AI will test what loyalty really means 

Personal AI can absorb much of the effort customers spend researching alternatives, comparing prices, negotiating, waiting and navigating complicated processes. Unlike people, it doesn’t get tired or frustrated enough to abandon the task. 

In doing so, personal AI will expose the friction that can masquerade as loyalty.  

It will also force us to reconsider how we measure a successful service experience. If two AI agents spend a few extra minutes reaching a better resolution without requiring the customer’s involvement, speed tells us very little. Successful task completion, resolution quality, customer effort and cost per outcome offer a better view of what was achieved. 

Genesys is ready to support these interactions today. Poppy creates an open, trusted way for personal AI to enter the enterprise with its identity and delegated authority understood. Genesys Cloud takes that request through to the appropriate API, workflow, enterprise AI or employee while preserving the context, controls and governance needed to deliver the customer’s authorized outcome. 

Personal AI has the potential to save consumers enormous amounts of time and reserve employees’ attention for moments where human judgment and connection are essential. 

Soon, the best customer service interaction may be the one the customer never has to experience. 

 

* Genesys Cloud Navigator is expected to be generally available in the fourth quarter of the company’s current fiscal year (Nov. 1, 2026-Jan. 31-2027), with Genesys Cloud Orchestrator expected to be generally available in the first quarter of company’s next fiscal year (Feb. 1-April 30, 2027).   

Forward-Looking Statements
Statements in this blog that are not historical or current facts are forward-looking statements that involve risks and uncertainties. Unless required by law, Genesys undertakes no obligation to update or revise any forward-looking statements to reflect circumstances or events after the date of this blog.