MANIFOLD
Will Anthropic, OpenAI, Deep-mind or Meta publish an app mainly for AI Agents by end of March 2026?
31
Ṁ1kṀ5.7k
Mar 31
56%
chance
12

  • Update 2026-02-04 (PST) (AI summary of creator comment): The app's main purpose should be for AI agents to interact with each other (e.g., to play games or similar activities), rather than primarily for agents to accomplish tasks for a specific user (even if multiple agents work together on those tasks).

  • Update 2026-02-04 (PST) (AI summary of creator comment): The main audience of the app must be agents, not humans. Examples include:

    • Agents interacting with each other

    • Agents playing games

    • Agents renting humans

A multi-agent coding app where agents work together to accomplish tasks for a specific human user would not qualify, as the main audience would still be humans.

  • Update 2026-03-09 (PST) (AI summary of creator comment): A qualifying app must have agents as the main audience, but humans can still benefit from the agents' work as a secondary effect. For example, a website where agents compete against each other, and human researchers find the interactions useful, would still count.

Market context
Get
Ṁ1,000
to start trading!
Sort by:
bought Ṁ500 YES

@Soli does a company acquiring an app mainly for AI agents count as publishing?

@Soli as in mostly for ai agents to interact with one another? would a coding app that is multi-agent count?

as in its main purpose is to accomplish tasks for a specific user but many ai agents work together on the same set of tasks

@Bayesian yeah mainly for agents to interact with each other or play games or whatever

@Bayesian yeah mainly for agents to interact with each other or play games or whatever or rent humans haha like the main audience of the app are agents and not humans

bought Ṁ500 NO

@Soli It sounds like a coding apop that happened to be multi-agent wouldn't have a "main audience of agents," but rather of humans still, so it would or wouldn't count? You said yeah but then it sounded like you disagreed it would count.

@Panfilo agree such a coding app wouldn't suffice to resolve the market yes, since the main audience of the app has to be agents, however, if humans benefit from the agents work but the main audience is agents it would still count. for example, a website where agents compete against each other, and human researchers find the interactions useful would still count towards resolving this market yes. Does this make sense?

sold Ṁ493 NO

@Soli No, and I'm withdrawing my mana due to the ambiguity. Nothing personal! The AI additions to the description might also be exacerbating things, ironically.

© Manifold Markets, Inc.TermsPrivacy