Open Agentic Web Challenge offers $13,000 for interoperable AI agents
A new global challenge is inviting developers, researchers, startups, businesses and students to build open, interconnected AI agents on today’s web. The Open Agentic Web Challenge offers a $13,000 prize fund and aims to test practical ways for agents to discover, trust and work with one another.
- $13,000 in prizes will recognize contributions to the challenge.
- Entries are open to developers, researchers, startups, businesses and students worldwide.
- The first contest runs on DevPost until 11th January 2027.
- The initiative builds on existing web technologies including HTML and JavaScript.
- Winners will be announced at the World Economic Forum Annual Meeting in Davos in January 2027.
Challenge targets a more open agent economy
Organizations spanning research, industry, technology standards and civil society announced the initiative at Web Summit in Lisbon on Nov. 12, 2026. The first contest is running on DevPost until 11th January 2027, with winners due to be announced at the World Economic Forum Annual Meeting in Davos in January 2027.
The challenge asks participants to create publicly accessible, interconnected agentic websites. Entrants can also develop technologies and standards that allow agents to discover one another and cooperate in ways that are safe, trusted and reliable. Additional organizations can join as collaborators.
According to the organizers, AI agents are beginning to discover services, communicate, negotiate and transact on behalf of people and organizations. However, the infrastructure for discovery, identity, trust and exchanging value is still taking shape.
Building on HTML, JavaScript and the existing web
The initiative presents an alternative to keeping personal AI agents confined to the platforms where they were created. Its proposed model would let people choose their agents, allow organizations to make services discoverable to agents, and enable independently developed agents to interact across platforms.
The challenge does not require an entirely new network. Instead, it builds on the existing web stack, including familiar technologies such as HTML and JavaScript. Projects may explore discovery, identity, multi-agent orchestration, portable agent programs, transactions and governance.
Participating organizations have contributed open-source or research starting points, although using them is not required. MIT Media Lab’s NANDA program offers WebAgentica and NandaIndex, described as research prototypes for signed, portable agent packages that can run in a user-controlled runtime such as OpenClaw or a browser. Cognizant’s Neuro® San is an open-source framework for building and orchestrating multi-agent systems. Mozilla’s Otari is an open-source LLM router designed to give agent applications flexible access to different models.
Backers span research, industry and standards
The initiative is supported by Cognizant AI Lab, MIT Media Lab’s NANDA program, Mozilla, United Nations ITU-T SG17, AWS, UT Austin School of Computing and AI Commons. Their stated contributions cover multi-agent systems, open-web principles, agent discovery, identity, trust, cloud infrastructure, academic research and the traceability of information used by agents.
ITU-T Study Group 17, described in the announcement as the United Nations’ lead group on security standards, is expected to bring expertise in digital identity and trust. The organizers say results from the challenge can help show what is feasible and deployable, while informing international standards work including ITU-T Study Group 17 and its Focus Group FG-TIDA.
For users and businesses, the project is testing whether agents can remain portable and user-controllable rather than being tied to a single platform. Builders can enter through the challenge’s DevPost site, while organizations seeking to collaborate can request more information at worldwideagenticweb.org.

