Fetch AI, a startup based and led by former DeepMind founding investor, Humayun Sheikh, at present introduced the discharge of three interconnected merchandise designed to supply the belief, coordination, and interoperability wanted for large-scale AI agent ecosystems.
The launch consists of ASI:One, a personal-AI orchestration platform; Fetch Enterprise, a verification and discovery portal for model brokers; and Agentverse, an open listing internet hosting greater than two million brokers.
Collectively, the system positions Fetch as an infrastructure supplier for what it calls the “Agentic Web”—a layer the place client AIs and model AIs collaborate to finish duties as a substitute of merely suggesting them.
The corporate says the instruments handle a central limitation in present client AI: fashions can present suggestions however can’t reliably execute multi-step actions that require coordination throughout companies. Fetch’s strategy facilities on enabling brokers from totally different organizations to interoperate securely, utilizing verified identities and shared context to finish end-to-end workflows.
“We’re creating the same foundation for agents that Google created for websites,” mentioned Humayun Sheikh, Founder and CEO of Fetch AI, and an early investor in DeepMind, in a press launch supplied to VentureBeat. “Instead of just finding information, your personal AI coordinates with verified brand agents to get things done.”
Background: Fetch’s Founding and DeepMind Connection
Fetch AI was based in 2017 by Humayun Sheikh, an entrepreneur whose early funding in DeepMind helped assist the corporate’s business improvement earlier than its acquisition by Google. “I was one of the first five people at DeepMind and its first investor. My check was the first one in,” Sheikh mentioned, reflecting on the interval when superior machine studying analysis was nonetheless largely inaccessible outdoors main expertise firms.
His early expertise helped form Fetch’s route. “Even in 2013, it was clear to me that agentic systems were going to be the ones that worked. That’s where I focused—on the agentic web,” Sheikh famous. Fetch constructed on this thesis by creating infrastructure for autonomous software program brokers, specializing in verifiable identification, safe knowledge alternate, and multi-agent coordination.
Over the previous a number of years, the corporate has expanded to a 70-person group throughout Cambridge and Menlo Park, raised roughly $60 million, and collected a couple of million customers interacting with its mannequin—knowledge that knowledgeable the design of the newly launched merchandise.
Sheikh added that his determination to bootstrap the corporate initially got here instantly from the proceeds of the DeepMind exit, noting within the interview that whereas the sale to Google was “a good exit,” he believed the group might have held out for a better valuation.
The early self-funding interval allowed Fetch to start work in 2015—effectively earlier than transformer architectures went mainstream—on the speculation that agentic infrastructure would develop into foundational to utilized AI.
ASI:One — A Platform for Multi-Agent Orchestration
On the core of the launch is ASI:One, a language mannequin interface designed particularly for coordinating a number of brokers somewhat than addressing remoted queries. Fetch describes it as an “intelligence layer” that handles context sharing, process routing, and choice modeling.
The system shops user-level indicators akin to favored airways, dietary constraints, price range ranges, loyalty program identifiers, and calendar availability. When a consumer requests a fancy process—akin to planning a visit with flights, inns, and restaurant reservations—ASI:One retrieves these preferences and delegates work to the suitable verified brokers. The brokers then return actionable outputs, together with stock and reserving choices, somewhat than generic suggestions.
In observe, ASI:One features as a workflow generator throughout organizational boundaries. Against this with typical LLM purposes, which frequently depend on APIs or RAG strategies to floor data, ASI:One is constructed to coordinate autonomous brokers that may full transactions. Fetch notes that personalization improves over time because the mannequin accumulates structured choice knowledge.
Sheikh emphasised the excellence between orchestrated execution and conventional AI output. “This isn’t searching for options separately and hoping they work together,” he mentioned. “It’s orchestration.”
He added that Fetch’s structure is deliberately modular: “Our architecture is a mix of agentic and expert models. One large model isn’t enough—you need specialists. That’s why we built ASI1, tuned specifically for agentic systems.”
The interview additionally revealed new particulars about ASI:One’s personalization methods: the platform makes use of a number of user-owned information graphs to retailer preferences, journey historical past, social connections, and contextual constraints.
These information graphs are siloed per consumer and never co-mingled with any Fetch-operated knowledge. Sheikh described this as a “deterministic backbone” that provides the non-public AI a secure reminiscence layer past the probabilistic output of a single giant mannequin.
ASI:One launches in Beta at present, with a broader launch deliberate for early 2026. Fetch additionally gives ASI:One Cellular, launched earlier this yr, giving customers entry to the identical agent-orchestration capabilities on iOS and Android. The cell app connects on to Agentverse and the consumer’s information graphs, enabling on-the-go process execution and real-time interplay with registered brokers.
Fetch Enterprise — Verified Id and Model Management
To allow dependable coordination between customers and corporations, Fetch is introducing a verification and discovery portal referred to as Fetch Enterprise.
The platform permits organizations to confirm their identification and declare an official Model Agent deal with—for instance, @Hilton or @Nike—no matter which instruments they use to construct the underlying agent.
Fetch positions the product as an analogue to ICANN area registration and SSL certificates methods for web sites. Verified standing is meant to guard customers from interacting with counterfeit or untrusted brokers, an issue the corporate describes as a significant barrier to widespread agent adoption.
The system consists of low-code instruments for small companies to create brokers in just a few steps and join real-time APIs akin to stock, reserving methods, or CRM platforms.
“With Fetch, you can create an agent in one minute. It gets a handle, like a Twitter username, and you can personalize it completely—even give it your social media permissions to post on your behalf,” Sheikh mentioned. As soon as a model claims its namespace, its agent turns into discoverable to client AIs and different brokers inside Agentverse.
The corporate has pre-reserved hundreds of name namespaces in anticipation of demand. Verification standing persists throughout any platform that integrates with Agentverse, creating a conveyable identification layer for enterprise brokers.
The interview highlighted that Fetch Enterprise inherits web-trust primitives instantly: area homeowners confirm their identification by inserting a brief code snippet into their present web site backend, permitting the system to cross a cryptographic problem and grant the agent an authenticity badge just like a “blue check” for agent identities. Sheikh framed this as “reusing the trust layer the web already spent decades building.”
Firms can start claiming brokers now at enterprise.fetch.ai.
Agentverse — An Open Listing of Extra Than Two Million Brokers
The ultimate element of the discharge is Agentverse, an open listing and cloud platform that hosts brokers and permits cross-ecosystem discoverability. Fetch states that thousands and thousands of brokers have already registered, spanning journey, retail, leisure, meals service, and enterprise classes.
Agentverse gives metadata, functionality descriptions, and routing logic that ASI:One makes use of to establish acceptable brokers for particular duties. It additionally helps safe communication and knowledge alternate between brokers. The corporate notes that the listing is platform-agnostic: brokers constructed with any framework can be a part of and interoperate.
In response to Sheikh, the dearth of a discovery layer is one cause most AI brokers see little or no utilization. “Ninety percent of AI agents never get used because there’s no discovery layer,” he mentioned.
He framed the function of Agentverse in additional technical phrases: “Right now, if you build an agent, there’s no universal way for others to discover it. That’s what AgentVerse solves—it’s like DNS for agents.” He additionally described the system as an integral part of the rising agent economic system: “Fetch is building the Google of agents. Just like websites needed search, agents need discovery, trust, and interaction—Fetch provides all of that.”
The interview additional underscored that Agentverse is cloud-agnostic by design. Sheikh contrasted this with competing agent ecosystems tied to particular cloud suppliers, arguing {that a} common registry is barely viable if unbiased of proprietary cloud environments. He mentioned the open structure permits an LLM to question any agent “within one minute of deployment,” turning agent publication right into a near-instantaneous course of just like registering a site.
Agentverse additionally integrates cost pathways, enabling brokers to execute purchases utilizing companions akin to Visa, Skyfire, and supported stablecoins. Customers can configure spending limits or require express approval for transactions.
Business Context and Implications
Fetch’s launch comes at a time when client AI platforms are exploring the shift from static chat interfaces towards autonomous brokers able to finishing actions. Nonetheless, most agent methods stay restricted by siloed architectures, restricted interoperability, and weak verification requirements.
Fetch positions its infrastructure as a response to those limitations by offering a cross-platform coordination layer, identification system, and listing service. The corporate argues that an agent ecosystem requires constant verification mechanisms to make sure that customers work together with genuine model representatives somewhat than imitations. By establishing namespace management and moveable belief indicators, Fetch Enterprise goals to fill a spot just like early internet area verification.
On the identical time, ASI:One makes an attempt to centralize consumer choice knowledge in a approach that allows extra environment friendly personalization and multi-agent coordination. This strategy differs from generalist LLM purposes, which frequently lack persistent choice architectures or direct entry to brand-controlled brokers.
The interview additionally made clear that micropayments and digital transaction infrastructure are central to Fetch’s long-term imaginative and prescient. Sheikh referenced integrations with protocols akin to Coinbase’s 402 and AP2, positioning these capabilities as important for autonomous brokers to finish end-to-end duties that embrace monetary execution.
Takeaway
Fetch’s mixed launch of ASI:One, Fetch Enterprise, and Agentverse introduces an interconnected stack designed to assist large-scale deployment and utilization of AI brokers. The corporate frames the system as foundational infrastructure for an agentic ecosystem, the place client AIs can coordinate with verified model brokers to finish duties reliably and securely. The additions to its identification, discovery, and orchestration layers replicate Fetch’s long-standing thesis—rooted partly in classes from DeepMind’s early improvement—that intelligence turns into significant solely when paired with the capability to behave.
