Introduction
Fetch AI’s latest announcement marks a pivotal moment in the evolution of consumer artificial intelligence. By launching three tightly coupled products—ASI:One, Fetch Business, and Agentverse— the company is attempting to solve a problem that has long plagued the AI ecosystem: the gap between recommendation and execution. Traditional language models excel at generating text and providing suggestions, yet they lack the mechanisms to coordinate real‑world actions across multiple service providers. Fetch’s vision is to create an “Agentic Web,” a layer where autonomous agents representing both consumers and brands collaborate to complete complex, multi‑step tasks. This ambition echoes the early days of the internet, when a simple search engine was insufficient to perform transactions; instead, a web of interconnected protocols and trust mechanisms was required. The new stack promises to bring that same level of infrastructure to the world of AI agents, offering a foundation that could enable everything from booking a flight to negotiating a business contract to ordering groceries—all through a single, intelligent interface.
The launch is not merely a product release; it is a statement about the direction of AI. By positioning itself as an infrastructure provider rather than a consumer‑facing application, Fetch signals that the next wave of AI will be built on shared standards, verifiable identities, and secure data exchange. In the following sections we will unpack how each component of the stack works, why they are necessary, and what implications they hold for businesses, developers, and end users.
Main Content
The Agentic Vision
The term “agentic” refers to systems that can act autonomously, making decisions and executing actions on behalf of a user. Fetch’s founder, Humayun Sheikh, has long argued that agentic systems are the future of AI. His early investment in DeepMind and his experience building autonomous software agents in the mid‑2010s gave him a unique perspective on the limitations of current models. He observed that while large language models can generate plausible text, they cannot reliably coordinate with external services to complete a task that requires multiple steps, such as booking a hotel, ordering a meal, and arranging transportation. The solution, he believes, lies in a distributed network of verified agents that can communicate securely and share context.
Fetch’s stack is designed to provide the three pillars that make this vision possible: orchestration, verification, and discovery. Orchestration is handled by ASI:One, a specialized language model that routes tasks to the appropriate agents. Verification is managed by Fetch Business, which gives brands a trusted identity and a mechanism to claim a unique namespace. Discovery is facilitated by Agentverse, an open directory that allows agents to be found and interacted with across the ecosystem. Together, these components create a self‑sustaining environment where consumer AIs can delegate work to brand agents, and brand agents can trust that they are interacting with legitimate users.
ASI:One – Orchestrating Multi‑Agent Workflows
ASI:One is the heart of the stack. Unlike conventional LLMs that answer isolated queries, ASI:One is engineered to coordinate multiple agents in a single workflow. It functions as an intelligence layer that stores user preferences—such as favored airlines, dietary restrictions, budget limits, loyalty program IDs, and calendar availability—in a user‑owned knowledge graph. When a user asks for a complex request, ASI:One retrieves the relevant preferences, determines which agents are best suited to handle each sub‑task, and delegates the work accordingly.
For example, if a user wants to plan a weekend getaway, ASI:One will first consult the user’s knowledge graph to identify preferred airlines and hotels. It will then route a flight‑booking request to a verified airline agent, a hotel‑reservation request to a hotel agent, and a restaurant‑booking request to a dining agent. Each agent returns actionable outputs—such as inventory, pricing, and booking confirmations—which ASI:One aggregates into a single, coherent plan. This orchestration eliminates the need for the user to manually coordinate with each service provider, reducing friction and the risk of errors.
ASI:One’s architecture is deliberately modular. Sheikh explains that a single large model is insufficient for the diversity of tasks that agents must perform. Instead, Fetch has built a mix of agentic and expert models, with ASI:One tuned specifically for agentic coordination. The platform also leverages deterministic knowledge graphs to provide a stable memory layer, ensuring that the AI’s understanding of user preferences remains consistent over time. This approach contrasts with the probabilistic nature of most LLMs, which can produce different outputs for the same prompt.
Fetch Business – Trust and Verification
The success of an agentic ecosystem hinges on trust. Without a reliable way to verify that an agent truly represents a brand, consumers risk falling prey to counterfeit or malicious actors. Fetch Business addresses this by offering a verification and discovery portal that mirrors the trust mechanisms of the web. Brands can claim a unique handle—such as @Hilton or @Nike—by inserting a short code snippet into their existing website backend. This snippet triggers a cryptographic challenge that proves ownership of the domain, after which the brand receives a verified badge.
Once verified, a brand’s agent becomes discoverable to consumer AIs and other agents within Agentverse. The verification status persists across any platform that integrates with Agentverse, creating a portable identity layer. This is analogous to how SSL certificates and domain registrations provide trust for websites. By reusing the web’s established trust primitives, Fetch Business reduces the friction for brands to enter the agentic space while simultaneously protecting consumers.
The platform also offers low‑code tools that allow small businesses to create agents quickly. These agents can connect to real‑time APIs—such as inventory systems, booking engines, or CRM platforms—enabling them to perform transactions on behalf of customers. The result is a seamless experience where a consumer’s personal AI can negotiate with a verified brand agent, complete a purchase, and receive confirmation, all without leaving the conversation.
Agentverse – The Universal Directory
Even the most sophisticated orchestration and verification systems would be useless without a way for agents to find one another. Agentverse serves as the universal directory, akin to DNS for the agentic world. It hosts metadata, capability descriptions, and routing logic for millions of agents spanning travel, retail, entertainment, food service, and enterprise categories.
Agents built with any framework can join Agentverse, making the directory truly platform‑agnostic. When ASI:One needs to delegate a task, it queries Agentverse to identify agents that match the required capabilities and have the necessary verifications. Agentverse also facilitates secure communication and data exchange between agents, ensuring that sensitive information is transmitted safely.
The directory’s design addresses a critical bottleneck in the current AI landscape: the lack of a discovery layer. Sheikh notes that ninety percent of AI agents never get used because there is no universal way for others to find them. By providing a searchable, verifiable registry, Agentverse transforms the agent economy from a fragmented set of isolated bots into a cohesive ecosystem.
Industry Implications and Future Outlook
Fetch’s integrated stack positions it as a foundational infrastructure for the next generation of AI applications. For businesses, the ability to claim a verified brand agent and connect it to a consumer‑facing AI opens new revenue streams and customer engagement models. For developers, the modular architecture and open directory reduce the barrier to entry, enabling rapid prototyping of agentic services.
The long‑term vision extends beyond simple booking or ordering. Fetch has hinted at integrating payment pathways, including partnerships with Visa, Skyfire, and stablecoin protocols. This would allow agents to execute purchases autonomously, subject to consumer‑defined limits or approval steps. Such capabilities could enable entirely new business models—think of an AI that negotiates a lease, signs a contract, and pays the vendor—all in a single interaction.
The broader AI community will likely watch closely as Fetch demonstrates the viability of a large‑scale, trustworthy agent ecosystem. If successful, the model could inspire similar infrastructure projects, leading to a more interconnected AI landscape where agents can collaborate across domains, share data securely, and deliver end‑to‑end services.
Conclusion
Fetch AI’s launch of ASI:One, Fetch Business, and Agentverse represents more than a product rollout; it is a bold step toward an agentic future where AI can act, not just advise. By combining orchestration, verification, and discovery into a single, interoperable stack, Fetch addresses the core limitations that have held back consumer AI from becoming truly transformative. The result is a platform that empowers brands to engage customers through verified agents, enables users to delegate complex tasks to a network of trusted services, and lays the groundwork for a new economy of autonomous agents. As the world moves toward more sophisticated AI interactions, infrastructure like Fetch’s will be essential to ensuring that these systems are secure, reliable, and scalable.
Call to Action
If you’re a developer, entrepreneur, or business leader interested in exploring the possibilities of autonomous agents, now is the time to get involved. Sign up for early access to ASI:One, claim your brand’s verified handle on Fetch Business, or explore the Agentverse directory to discover existing agents that can be integrated into your workflows. By participating in this emerging ecosystem, you can help shape the standards, protocols, and best practices that will define the next generation of AI‑powered services. Join the conversation, experiment with the tools, and be part of the movement that turns AI from a recommendation engine into a full‑fledged partner capable of getting things done.