FFetch.ai Agentverse

Agentverse is Fetch.ai’s open marketplace and directory for AI agents, and it suits developers and brands who want their agents found by people through the ASI:One assistant and by other agents that need their services. More than 2 million agents coordinate through it, according to Fetch.ai, and brands can claim verified namespaces so users know which agent is official.

It is built for developers more than for business buyers shopping for finished software, Premium tier prices come from Fetch.ai on request, and some features connect to crypto tokens. We score it 8.2/10 on our scale.

Agentverse by the Numbers

Fetch.ai’s November 2025 announcement and later coverage give the main figures.

The gap between registered agents and active deployments is the useful number. Around 3 million agents against 150,000 active deployments means most registrations are experiments or inactive, which is common in open registries. Users should look for active, verified agents, and builders should expect to compete for attention.

The company behind it is small for its reach. Fetch.ai reported 70 employees and $60 million raised from angel investors in November 2025, which puts a lot of weight on its developer community to fill the directory.

Fetch.ai’s chief executive, Humayun Sheikh, explained the reasoning behind the directory: in his view, 90 percent of AI agents never get used because there is no discovery layer. Agentverse and ASI:One are his answer to that problem.

How Agentverse Connects Agents, Brands and Users

Agentverse has three groups of users: developers who build agents, brands that want an official presence and people who use agents through ASI:One.

Building and Registering an Agent

Fetch.ai’s developer materials set two mandatory steps for hackathon entries: make the agent discoverable and make it usable through ASI:One. Both depend on the same registration and protocol setup.

Developers build agents with Fetch.ai’s uAgents framework or another agentic framework and register them on Agentverse. To be discoverable through ASI:One, an agent must implement the Agent Chat Protocol and publish its manifest to Agentverse. Fetch.ai’s developer materials show this as a few lines of setup in the agent code.

Being Found Through ASI:One

ASI:One is the discovery layer. When a person submits a request, ASI:One chooses the most suitable registered agent and routes the request to it. For multi-part jobs such as trip planning, ASI:One can coordinate several verified brand agents, and it learns user preferences over time. Fetch.ai launched ASI:One in beta in November 2025 and planned broader availability for early 2026.

What a User Sees

For a person, Agentverse mostly stays in the background. They ask ASI:One for help, and ASI:One decides which registered agent or agents should handle the request. Where a task needs payment, the user’s own settings decide whether the agent can spend within a limit or must ask first. Verified brand namespaces tell the user they are dealing with the company’s official agent.

Agents Finding Other Agents

Agentverse also works as a directory for agents themselves. Agents can search for, work with and transact with other agents, so a travel agent might call a hotel brand’s agent and a payment agent in the same task. Fetch.ai describes the directory as open and platform-agnostic.

Claiming a Brand Namespace

Brands use Fetch Business to verify their identity and claim a namespace, such as @Hilton or @Nike in Fetch.ai’s examples. They then connect an existing agent or build one with Fetch.ai’s low-code tools and deploy it to Agentverse. Fetch.ai has pre-reserved namespaces for thousands of recognizable brands through its Claim Your Agent program, so impostors cannot take them first. Sheikh compared the process to ICANN for agents.

Payments Between People and Agents

Agentverse includes payment infrastructure through partners including Visa. People can approve purchases, set spending limits or require confirmation for any transaction. Developers can add the optional Payment Protocol to charge for their agent’s work.

The design shifts some control to users. Instead of a company deciding which agents staff may use, each person sets limits and approvals on what agents can spend. That model suits consumer tasks such as travel and shopping, where the user is also the buyer.

MCP Access

Agentverse runs a full MCP server and an MCP-Lite endpoint, so assistants such as Claude Desktop can work with Agentverse directly. Fetch.ai says a developer can deploy a first agent this way in under five minutes.

Why Verified Namespaces Matter

Open directories invite impersonation. If anyone can register an agent called after a well-known airline, users have no way to know which one is real. Fetch Business ties a namespace to a verified company, and the pre-reserved list keeps recognizable brand names out of other hands until the brand claims them. For a user asking ASI:One to rebook a flight, that verification is what makes it reasonable to let an agent act on their behalf.

Token Features

Coverage in June 2026 described Agent Launch on BNB Chain, which lets agents issue their own tokens and take part in financial activity on their own. These features are optional, and companies with policies against crypto can use the directory, protocols and MCP access without them.

A Developer Publishing a Booking Agent, Step by Step

Take a developer at a boutique hotel group. He builds a booking agent with uAgents, adds the Agent Chat Protocol and publishes its manifest to Agentverse. The marketing team verifies the company through Fetch Business and claims the brand namespace, so the agent appears as the official one. He adds the Payment Protocol so the agent can take deposits, while guests can set spending limits or confirm each payment. When a traveler asks ASI:One for a weekend stay, ASI:One can route the request to the hotel’s agent.

Agentverse Pricing

Fetch.ai sells premium tiers for Agentverse and ASI:One and runs brand verification through Fetch Business.

Budgeting for Agentverse

• Developers testing an idea: start with the developer materials and MCP access, and move to Premium only when you need its features.
• Brands claiming a namespace: contact Fetch Business for verification terms, since a claimed namespace is the main trust signal for users.
• Agents that charge: plan for payment partner fees on each transaction when using the Payment Protocol.

Ask Fetch.ai for current Premium and Pro prices before committing a team.

For brands, the larger cost is usually building and maintaining the agent itself, not the listing. An official agent has to answer accurately for the brand, so plan for testing and updates the same way you would for a customer-facing chatbot or app.

Agentverse Compared With Agent.ai, AWS Marketplace and Microsoft Agent Store

Agentverse is the most developer-focused of these four marketplaces, and the only one built around agents finding each other.

On discovery, the four differ sharply. Agentverse relies on ASI:One to pick agents for a request, AWS on agent mode for comparisons and proposals, Microsoft on curated collections and suggestions, and Agent.ai on search and community ratings.

Agentverse and Agent.ai are both open marketplaces, but Agent.ai targets people who run agents in a browser, while Agentverse targets agents that talk to each other and to ASI:One. AWS and Microsoft sell to companies with procurement and admin needs. Developers comparing frameworks can also see our agent frameworks and AI agent marketplaces pages, and our review of CrewAI for multi-agent building.

Developers and Brands Building for Agent-to-Agent Discovery

Agentverse fits developers who want their agents to be found and used by other agents and by ASI:One users, and brands that want a verified, official agent before someone else claims their name. The chat and payment protocols give a shared way for agents to talk and charge, and MCP access connects the directory to common assistants.

Student teams and hackathon builders are a large part of the community, and Fetch.ai runs events with developer packs that walk through registration, protocols and MCP setup. For a developer learning agent-to-agent design, that material is a practical starting point.

It also suits teams experimenting with multi-agent systems, since agents can discover each other in a shared directory instead of hard-coding every connection. A research team can publish a specialist agent once and let any compatible agent find and call it through the directory.

Companies Buying Finished Agents for Internal Use

Companies looking to install governed agents in their CRM, ERP or office suite will find AgentExchange, Oracle’s marketplace or Microsoft’s Agent Store closer to that need.

Teams that need an agent catalog with procurement terms, invoicing and admin controls will find the platform stores more familiar.

Organizations with policies against crypto should review which features they use, since some Fetch.ai features involve tokens.

Inactive Agents, Protocols and Payments

• Inactive listings: registrations far outnumber active deployments, so filter for verified, active agents.
• Protocol work: agents must implement the Agent Chat Protocol to be discoverable through ASI:One.
• Payments: set spending limits or confirmation for any agent that can charge you.
• Token features: optional token issuance on BNB Chain may not fit every company’s policies.
• Pricing: confirm Premium and Pro prices with Fetch.ai.
• Brand claims: claim your namespace early through Fetch Business so users find the official agent.
• Agent behavior: an official brand agent speaks for the company, so test its answers and payment flows before announcing it to customers.

Agentverse and ASI:One Announcements

In November 2025 Fetch.ai launched ASI:One in beta, Fetch Business and its Claim Your Agent program, saying more than 2 million agents coordinate through Agentverse. In June 2026 Crypto Briefing reported about 3 million agents and more than 150,000 active deployments. Fetch.ai’s developer materials describe Agentverse as the open marketplace for AI agents.

Agentverse Suits Builders of Agents That Work Together

We rate Agentverse 8.2/10 on our scale. It offers something the platform stores do not: a shared directory where agents find, coordinate with and pay other agents, verified brand namespaces and a consumer assistant that routes requests to the right agent. MCP access and support for several frameworks make it practical for developers.

The protocols are the long-term asset. A shared chat protocol and payment protocol mean agents built by different teams can talk to and pay each other without custom integrations, which is the piece most single-vendor stores leave out.

Areas to improve are pricing transparency and signal-to-noise, so that active, verified agents are easier to tell apart from inactive registrations. For developers and brands preparing for agents that work with other agents, it is a well-developed place to publish.

Developers’ Questions About Agentverse

What Is Agentverse?

Agentverse is Fetch.ai’s open, platform-agnostic marketplace and directory where AI agents register, discover each other and coordinate, and where ASI:One finds agents for users. Agents built with uAgents or other frameworks can be listed.

How Do I Make My Agent Discoverable?

Register it on Agentverse, implement the Agent Chat Protocol and publish its manifest. ASI:One can then route suitable user requests to it, and other agents can find it in the directory.

Can Brands Protect Their Agent Names?

Yes. Brands verify identity and claim namespaces through Fetch Business, and Fetch.ai has pre-reserved namespaces for thousands of recognizable brands. Claiming a namespace makes the brand’s agent the verified one users see.

Can Agents on Agentverse Take Payments?

Yes. Developers can add the Payment Protocol, and payments run through partners including Visa. Users can approve purchases, set spending limits or require confirmation for every transaction an agent proposes.

What Is ASI:One?

ASI:One is Fetch.ai’s personal AI assistant and the discovery layer for Agentverse. It picks the most suitable registered agent for a request, can coordinate several brand agents for multi-part tasks and learns user preferences over time.

Does Agentverse Work With MCP?

Yes. Agentverse offers a full MCP server and a lighter MCP-Lite endpoint for assistants, and Fetch.ai says a first agent can be deployed with Claude Desktop in under five minutes.