Fetch AI: Pioneering the Future of Autonomous AI Agent Ecosystems
Founded by Humayun Sheikh, an early investor and founding member of DeepMind, Fetch AI introduces a groundbreaking suite of three integrated products designed to establish trust, seamless coordination, and interoperability within expansive AI agent networks. This innovative launch includes ASI:One, a personal AI orchestration platform; Fetch Business, a verification and discovery portal for brand agents; and Agentverse, an open directory hosting over two million AI agents.
Together, these offerings position Fetch AI as a critical infrastructure provider for what it terms the “Agentic Web”-a dynamic environment where consumer and brand AI agents collaborate to execute complex tasks, moving beyond mere suggestions to actual task completion.
From DeepMind Roots to Agentic Web Vision
Humayun Sheikh’s journey began with his pivotal role as one of the first five individuals at DeepMind and its initial investor, providing crucial support during the company’s formative years before its acquisition by Google. Recognizing early on the transformative potential of agentic systems, Sheikh envisioned a future where autonomous agents would drive AI applications. This foresight shaped Fetch AI’s mission to develop foundational infrastructure enabling secure identity verification, data exchange, and multi-agent collaboration.
Since its inception in 2017, Fetch AI has grown into a 70-member team spread across Cambridge and Menlo Park, securing approximately $60 million in funding and engaging over one million users. This extensive user interaction has informed the design and functionality of its latest product suite. Notably, the company’s initial self-funding phase, fueled by proceeds from the DeepMind exit, allowed it to pioneer agentic infrastructure development well before transformer models became mainstream.
ASI:One – Revolutionizing Multi-Agent Coordination
At the heart of Fetch AI’s ecosystem lies ASI:One, a specialized language model interface crafted to orchestrate multiple autonomous agents simultaneously rather than responding to isolated queries. Acting as an “intelligence layer,” ASI:One manages context sharing, task delegation, and user preference modeling to streamline complex workflows.
ASI:One maintains detailed user profiles, including preferred airlines, dietary restrictions, budget limits, loyalty memberships, and calendar availability. For example, when a user requests a comprehensive travel itinerary involving flights, accommodations, and dining reservations, ASI:One intelligently assigns tasks to verified agents who provide actionable options such as booking availability and pricing, rather than generic suggestions.
This platform transcends traditional large language model (LLM) applications by enabling autonomous agents to complete transactions across organizational boundaries. Personalization improves continuously as ASI:One accumulates structured preference data, creating a more tailored user experience over time.
Sheikh highlights the distinction between mere information retrieval and true orchestration: “ASI:One doesn’t just gather options separately and hope they align-it actively coordinates agents to execute tasks.” The architecture combines agentic and expert models, recognizing that a single large model cannot fulfill all specialized functions.
Additionally, ASI:One employs multiple user-owned knowledge graphs to securely store preferences, travel history, social connections, and contextual constraints. These graphs remain isolated per user, ensuring privacy and providing a deterministic memory layer that complements the probabilistic nature of LLM outputs.
Currently in Beta with a full release anticipated in early 2026, ASI:One is also available on mobile platforms, enabling users to orchestrate tasks and interact with agents on iOS and Android devices through direct integration with Agentverse and personal knowledge graphs.
Fetch Business: Establishing Verified Brand Identities
To foster trustworthy interactions between consumers and companies, Fetch AI introduces Fetch Business, a verification and discovery portal that allows organizations to authenticate their identity and claim official Brand Agent handles-such as @Marriott or @Adidas-regardless of the underlying agent technology.
Functioning similarly to domain registration and SSL certification systems on the web, Fetch Business aims to protect users from counterfeit or fraudulent agents, a significant obstacle to widespread adoption. The platform offers low-code tools enabling small businesses to create agents quickly and connect real-time APIs for inventory, booking, or customer relationship management.
Sheikh explains, “With Fetch, creating an agent takes just a minute. It receives a unique handle, akin to a social media username, and can be fully customized-even granted permissions to post on your behalf.” Once a brand secures its namespace, its agent becomes discoverable to consumer AIs and other agents within Agentverse.
Thousands of brand namespaces have been pre-reserved to meet anticipated demand. Verification persists across all platforms integrated with Agentverse, providing a portable and reliable identity layer for business agents. Verification is achieved by embedding a cryptographic challenge code into the brand’s existing website backend, granting a “blue check” authenticity badge that leverages decades of web trust infrastructure.
Agentverse: The Universal Directory for AI Agents
The third pillar of Fetch AI’s ecosystem, Agentverse, is an open, cloud-agnostic directory hosting over two million registered agents across sectors such as travel, retail, entertainment, food service, and enterprise solutions.
Agentverse supplies metadata, capability descriptions, and routing logic that ASI:One utilizes to identify and engage the most suitable agents for specific tasks. It also facilitates secure communication and data exchange between agents, regardless of the development framework used.
Sheikh points out a critical industry challenge: “Ninety percent of AI agents remain unused due to the absence of a discovery mechanism.” Agentverse addresses this by acting as a “DNS for agents,” enabling universal discoverability and interaction. This open architecture contrasts with competing ecosystems tied to proprietary cloud providers, ensuring rapid agent publication and interoperability.
Moreover, Agentverse integrates payment solutions through partners like Visa, Skyfire, and stablecoin platforms, allowing agents to execute financial transactions with configurable spending limits and user approvals, essential for autonomous end-to-end task completion.
Implications for the AI Industry and Future Outlook
Fetch AI’s comprehensive launch arrives amid a broader industry shift from static chatbots to autonomous agents capable of executing complex, multi-step actions. Current limitations in agent interoperability, verification, and siloed architectures hinder widespread adoption.
By delivering a unified coordination layer, robust identity verification, and an open discovery platform, Fetch AI addresses these challenges head-on. The company’s approach to persistent user preference management and direct access to verified brand agents marks a departure from generalist LLM applications, promising enhanced personalization and reliability.
Furthermore, the integration of micropayment and digital transaction protocols underscores Fetch’s vision for a fully autonomous agent economy, where AI agents can independently manage financial operations as part of their workflows.
In sum, Fetch AI’s interconnected stack-comprising ASI:One, Fetch Business, and Agentverse-lays the groundwork for scalable, secure, and efficient AI agent ecosystems. Rooted in lessons from DeepMind’s pioneering days, Fetch’s platform embodies the principle that intelligence gains true value only when paired with the ability to act decisively and reliably.
