Genesys Takes Action on AI Customers, Partners with Meta, Sierra to Deliver a Personal Agent Protocol

Earlier this week, Genesys shared its involvement in developing a new Personal Agent Protocol.

First announced by Meta and conversational AI giant Sierra, the protocol is an open standard that will define how personal agents interact with businesses.

Genesys is working closely with the two companies and several other partners to deliver this landmark innovation, with the v0.1 specification expected later this month. 

Those other partners include Rocket, Shopify, Stripe, Walmart, and Instinct (more on them later!).

But first, what are personal agents, how will this protocol work, and what should contact center leaders take away from the news?

What Are Personal Agents?

A personal agent is an AI assistant that works on the user’s behalf to complete a task. 

While many tasks will be personal errands, consumers are also starting to use these agents to book trips, fill forms, purchase goods, and even call contact centers on their behalf. 

Now, this isn’t anything new. Google Duplex gave the contact center industry a taste of personal AI assistants that could call businesses to schedule appointments, make reservations, and check their holiday opening hours. 

What is new is their sophistication.

Instinct is an excellent example. It’s a personal agent that can call businesses for customers, connect with their apps and devices, and take instructions from a text or WhatsApp message.

Now, consumer tech giants are getting in on the act, pushing this technology from the periphery and into the mainstream. 

What Impact Will They Have on Contact Centers? 

The launch of Meta Muse and OpenAI’s Dots agent proved to be watershed moments.

Integrated with WhatsApp, Muse sends emails, books travel, makes purchases, and will soon make calls, a feature already in public beta in North America.

Only three weeks after its launch last month, Muse was downloaded five million times.

But don’t be fooled into thinking that this is a flash in the pan. Expect personal agents to soon be available through the AI assistants embedded into smartphones, like Siri and Gemini.

When this (probably) happens, the personal agent will follow the customer’s voice instructions to interact with businesses on their behalf. 

That’s the trend line, and as AI agents become more accessible, contact centers can anticipate an explosion in AI customers calling in over the coming months.

As this happens, contact centers must prepare to handle four new modes of conversation:

1. Human customer to human support rep
2. Human customer to AI support rep
3. AI customer to human support rep
4. AI customer to AI support rep

That is a lot to process for contact center operations, IT, and security leaders.

Recognizing this, Genesys accepted the opportunity to partner with Sierra and Meta, helping develop the new Personal Agent Protocol and simplify part of the transition.

How Will the New Personal Agent Protocol Work? 

Personal agents will get stuck while attempting to complete tasks online and escalate to the company's support team.

This creates an immediate challenge for the organization: it may not know who the AI agent represents, what it is trying to accomplish, or what the customer has authorized it to do.

The Personal Agent Protocol is being designed to address this by handling authentication, empowering consumers, and giving companies visibility into what personal agents do through their websites, APIs, or company agents. It’s open for anyone to implement.

Under the proposed approach, when account access is needed, the customer can sign in on the company’s page or use credentials already configured with their personal agent. The customer decides whether to grant read-only or write access.

The proposed session model is built on OAuth, an established authorization standard, and maintains continuity across channels.

Meanwhile, the company determines which routes are available: its website, its APIs, or its own agent. Customers control the access they grant, while businesses set the parameters for what personal agents can do.

3 Considerations for Contact Center Leaders

“The Personal Agent Protocol is an important step toward helping businesses recognize personal AI agents acting on customers’ behalf and manage their authorized access. Yet, service leaders still have much more to consider,” said Charanya “CK” Kannan, SVP & Head of Product Marketing at Genesys.

Here are three considerations Kannan recommends for support leaders moving forward.

1. The Single-Designed Customer Journey Is Dead

For years, many contact centers focused on orchestrating one ideal experience for every customer who calls in with a particular intent. That thinking is now dead in the water. 

When a brand interacts with both human and AI customers, it may have to orchestrate them and pivot the conversation differently.

As a result, brands must think much harder about orchestration. 

Orchestration is also critical when AI-to-AI interactions get stuck, or neither agent knows how to end the call (yes, this happens!). Without this, these interactions can continue unchecked, sending AI costs skyrocketing.

2. Many Conventional Contact Center Metrics Just Won’t Make Sense Anymore 

When the customer is an AI agent, contact center metrics - from wait time and average handling time to sentiment scores - become largely irrelevant. 

“AI agents don't mind waiting on the line for 30 minutes. They don't care for empathy either. Instead, they just want a resolution.”

Charanya “CK” Kannan

SVP & Head of Product Marketing at Genesys

Given this, contact centers must be able to isolate what matters across the four modes of conversation mentioned earlier and track those metrics accordingly.

Those that apply human-centric metrics to conversations with AI customers will waste time measuring the wrong things, and potentially reward the wrong behaviors.

3. AI Accountability Could Get Messy 

As personal agents become more prevalent, brands will understand the significance of supporting them. After all, if they don't, a competitor might. 

That said, contact centers should be selective about which intents they support for personal AI agents, given the many unknowns around what happens when something goes wrong.

For instance, what if a personal agent makes an important commitment that their “owner” didn’t authorize? Alternatively, what if they take an action that leads to a bad outcome based on incorrect information given by the company’s AI agent?

Liability frameworks are still far from providing definitive answers to these questions.

There’s Lots More to Consider… 

While Kannan raises three key points, she notes that there is still much more to consider.

Take vulnerable customers, for example. A personal agent that can communicate on their behalf may seem appealing, but what happens if they become too reliant on an intermediary that doesn’t fully understand their situation?

This highlights the importance of journey orchestration that treats each customer as an individual, determining whether a personal AI agent is appropriate not just based on intent, but on the customer’s specific circumstances.

Genesys is already considering this future, expanding its orchestration capabilities at its recent Xperience 2026 event with the release of Navigator.

Catch up on that and all the other event announcements here.