Article

November 17, 2026

Do You Know Your Agent? For Most, the Phone Channel Is Still the Wild West

Know Your Agent (KYA) verifies AI agents online, but the phone channel doesn’t have a clear protocol yet.

Our call volume has skyrocketed. Our customer base hasn’t.

We’ve heard some version of that again and again over the past few weeks, especially since personal AI agents like Meta’s Muse started making outbound calls. In fact, Pindrop researchers have now identified more than 1.88 million confirmed AI agent interactions.1

These AI agents are making calls on behalf of your customers to check balances, file claims, and dispute charges. But your contact center wasn’t designed for this new type of caller. Handling them well starts with a framework that the rest of the industry is already starting to adopt: Know Your Agent.

What is Know Your Agent?

Know Your Agent (KYA) is the practice of detecting an AI agent, determining who sent it, and what it’s allowed to do. KYA adapts Know Your Customer (KYC) rules for the bot caller. 

At the core of KYA are three questions:

  • Is this an AI agent? 
  • Who does it represent?
  • What should it be allowed to do?

KYA also requires ongoing verification, not just a one-time check. Because an agent can drift, verification has to happen throughout the interaction.

Why today’s KYA protocols stop short of the phone

The industry is already working on KYA, but almost entirely for digital channels: your website or app, self-serve chatbots. Companies are partnering up to define personal agent protocols, and payment networks have taken steps to standardize authorization for agents.

Visa and Mastercard rely on Cloudflare’s Web Bot Auth, which uses HTTP Message Signatures. The agent signs each request it sends, and the merchant checks that signature before anything happens. 

Poppy, a personal agent protocol in development by Sierra AI, works the same way for customer service. Agents can come in through a company’s website, its API, or its own AI agent, and the customer provides permission to the agent by signing in.

Poppy also steers agents away from the phone channel, emphasizing the importance of letting agents accomplish tasks without needing to call. The protocol is vital in keeping everyday, simple tasks out of your call queue. 

But what happens to the bots that still get through? The phone will always be a backup when digital tasks fail, a bad actor’s agent won’t announce itself, and some agents will just choose to dial. For those callers, the protocol is still being developed.  

The agentic caller is already on the line

The rise in agentic callers is happening now. The 1.88M confirmed AI agent interactions happened in just the last several months, starting in July 2026.

On the risk side, Pindrop research found AI-powered attacks rose nearly 2,310% since late 2024.² We’re seeing these AI callers in healthcare, financial services, insurance, and retail. They probe for account information, eat up agent capacity, and pose as real customers.

There is an obvious fix: hang up on all bots. That works on bad actors, but it also hangs up on your customers’ assistants, and those customers end up back in your queue, waiting and annoyed.

What you need instead is a way to tell agents apart: whether an agent is calling, who it’s calling for, and what it should be allowed to do. In other words, you need KYA on the phone. 

What KYA looks like on the phone

Start with the simplest question: is this a machine? Pindrop BotStopper™ technology is built for that. It flags a machine caller by detecting synthetic speech, and checks the voice against a registry of 5,000+ known AI voices in the AI Voice Consortium. 

The questions that follow, which agent is it, who does it represent and what should it be allowed to do, are where Pindrop is taking the platform next. 

You don’t need the full picture to get started. Start with detection. 

Know when bots call

The future of the contact center will include bot callers. Your customers are sending them on purpose for legitimate reasons. And bad actors have already adopted agentic technology to scale their attacks. Before creating a robust policy, you need to know if you’re even talking to a bot. That starts with detection—with BotStopper. See how BotStopper works.

FAQs

What is Know Your Agent (KYA)?

Know Your Agent (KYA) is a framework of verifying an AI agent before and while it acts for someone. It asks three questions: is this an agent, who does it represent, and what is it allowed to do. It borrows the logic of Know Your Customer (KYC) and applies it to callers that aren’t human.

Can contact centers verify AI agents that call on a customer’s behalf?

Partly. Contact centers can detect when a caller is an AI agent. Confirming who sent the agent and what it’s allowed to do is still early on the phone, because calls don’t carry the signed credentials digital protocols use.

Sources

Don't assume. Verify.

Take the first step toward a safer, more secure future for your business.