---
title: "AI Agents Need Names Too"
date: 2026-10-03
author: "Erwin Groen"
tags:
  - Handshake
  - AI
  - AI Agents
  - Digital Identity
  - Headless Domains
  - HNS
description: "AI agents are beginning to get names and identities of their own. As they start acting on our behalf, the question is: what will we name them, and who owns their identity?"
image: "/media/ai-agents-need-names-too-v2.png"
---

As AI moves from answering questions to doing things on our behalf, agents will need identities that people, websites, services and other agents can recognise.

## Agents Are Becoming Individuals

AI companies are already beginning to give individual agents their own identities.

OpenAI has introduced **dots** individual representations for agents that can be named and recognised.

Meta has taken a similarly personal approach with **Muse**, represented by its character **Jolly**.

Giving an agent a name, colour or character makes it easier for us to recognise and interact with it.

## A Face Still Needs an Identity

Making AI agents look friendly and approachable makes sense.

These agents will eventually do much more than talk to us.

⚪️ They could send an email.

⚪️ Make a booking.

⚪️ Access an account.

⚪️ Research something.

⚪️ Communicate with another agent.

⚪️ Take actions on our behalf.

As AI agents become more capable, accountability becomes more important.

**⚪️ Which agent performed the action?**

**⚪️ Who controls it?**

**⚪️ Who authorised it?**

**⚪️ Can its identity be verified?**

A character or coloured dot may help us recognise an agent.

Accountability may require something more permanent.

## A Name Could Become Part of Accountability

Today we usually think about domain names as addresses for websites.

Tomorrow they could also help identify machines.

An AI agent with a persistent name could potentially carry an identity that other systems can recognise and verify.

That identity could tell another agent or service:

This is who I am.

This is what I am allowed to do.

This is where I can be reached.

**This action came from me.**

Names could therefore become part of the accountability layer of an agent driven internet.

## Platform Name or Owned Name?

What happens to your agent's identity if it exists entirely inside one company's platform?

You give your agent a name.

You personalise it.

Over time, that identity could become familiar to you and perhaps important to the services it interacts with.

But does that mean you actually **own the name?**

Or is the identity ultimately controlled by the platform providing the agent?

This is where the distinction between a **platform assigned identity** and an **independently controlled name** becomes interesting.

## Headless Domains Is Already Exploring This

**HeadlessDomains.com** is already exploring persistent names for AI agents using Handshake namespaces such as `.agent` and `.chatbot`.

Rather than designing domains primarily around websites for people to visit, the idea is designed around agents themselves.

An agent can have a persistent name together with machine readable information describing its identity, capabilities and services.

In other words, a name doesn't necessarily have to lead to a traditional website.

It can also help machines discover and identify each other.

That is a very different way of thinking about a domain name.

## Handshake Adds Another Question

Handshake takes the idea one step further because its names can exist independently of a single platform.

That opens up the possibility of an agent identity that isn't simply a username allocated by one company.

The underlying name can be held and controlled by its owner.

That doesn't solve every question around AI identity, security or accountability.

But it introduces an interesting principle:

**Your agent could have a name that you control.**

And potentially take that identity with it as technology and platforms change.

## AI May Make Names More Important

For years, there has been discussion about whether AI will make domain names less important.

If an AI agent finds information for us, perhaps fewer people will type addresses into browsers.

The internet will soon contain enormous numbers of agents, services and machines communicating directly with one another.

They will still need ways to identify who they are talking to.

They will still need trusted endpoints.

And increasingly, they will need persistent identities.

The audience for a name may no longer be just humans.

It could be machines as well.

## Who Should Own the Name?

We are still at the beginning of the agent era.

OpenAI has its dots.

Meta has Muse and Jolly.

Other companies will create their own versions.

Giving these agents friendly identities may make them easier for us to understand.

But as they become more capable, the question may not be what they look like.

It may be who controls their identity.

Because if AI agents are going to act for us, communicate for us and increasingly represent us online, their names will become important.

**AI agents need names too.**

The next question is who should own them. ⚪️

**Own The Name | Own The Rules | Own The Dot.**