#USA #Lisbon #AI Agents #Open Standards #Agentic Web

The Open Agentic Web Challenge: Building an Inclusive AI Economy

In a significant announcement at the Web Summit, a collaboration between organizations from research, industry, tech standards, and civil society unveiled the Open Agentic Web Challenge. The initiative aims to cultivate an open, secure, and interoperable network of AI agents. With a prize pool of $13,000, this competition invites developers, researchers, startups, businesses, and students to create publicly accessible, interconnected AI agents capable of working in synergy across multiple platforms.

Why Does This Matter?

As AI agents begin to routinely act on behalf of individuals and organizations—gathering information, negotiating deals, and facilitating transactions—the infrastructure that allows these agents to locate one another, build trust, and exchange value is still emerging. Currently, personal AI agents are largely confined to the platforms upon which they were developed. This lack of open standards risks increasing isolation, making it ever more challenging for individuals and businesses to connect agents with the services they traverse.

The challenge seeks to explore an alternative future where individuals can select their AI agents freely, organizations can make their services discoverable, and independently developed agents can communicate across various platforms. It endeavors to leverage existing web technologies like HTML and JavaScript to establish practical approaches for locating agents, identity management, orchestration of multiple agents, transferable agent programs, transactions, and governance.

Participation Details

The competition encourages participants to develop and test open components of the agent web using current web technologies. Developers can either create new components or leverage existing research projects and open-source solutions provided by participating organizations, including:

• - NANDA Web Agentica and NandaIndex from MIT Media Lab that connects resources allowing agents to run in user-controlled environments.
• - Neuro® San from Cognizant, a framework enabling the creation and orchestration of multi-agent systems that allows developers to coordinate cooperating agents.
• - Otari from Mozilla, an open-source LLM router for flexible model access tailored for agent applications.

While utilizing these contributions is not mandatory for competition entry, they will address various aspects of the challenge and enhance collaborative efforts.

Expert Opinions on the Challenge

The competing organizations shared their insights regarding the initiative:

• - Babak Hodjat (Chief AI Officer at Cognizant) emphasized, "Personal agents must work for you and communicate freely across platforms. The open agent economy empowers everyone to innovate."
• - Ramesh Raskar (Associate Professor at MIT Media Lab) noted, "Our research is about ensuring agent programs are transferable and controllable, facilitating inter-agent communication and connection to information."
• - Raffi Krikorian (CTO of Mozilla) conveyed excitement about providing developers with open-source tools to oversee their AI technology stack's implementation.
• - Peter Stone (head of the School of Computing at UT Austin) mentioned how this initiative provides a pathway for users to manage their interactions with AI agents across different service providers.
• - Madisen Taylor (director of AI Commons) discussed the importance of making agents accountable for their information sources, ensuring increased access and trust for users.

The Broader Impact

This initiative brings together diverse entities from the business community, academic institutions, civil society, and the international policy landscape. By promoting collaborative standards, the Open Agentic Web Challenge aspires to build a global environment where AI agents thrive while upholding user autonomy and choice.
Participants interested in joining this initiative or learning more can explore further information through the worldwideagenticweb.org website. As we approach the deadline for submissions, the anticipation continues to build around the innovations that this challenge could yield in the landscape of AI technology.

In conclusion, the Open Agentic Web Challenge not only presents a thrilling competition with attractive rewards but also holds the potential to redefine how we interact with technology, highlighting the need for an accessible and equitable web for all users and creators.