Show HN: An AI agent runs ten web errands a day, every transcript published (agentability.org)

A new open experiment named "Agentability" has emerged, showcasing an AI agent that completes ten real-world web errands daily by utilizing public web searches. The AI operates without any human intervention, login requirements, or JavaScript, simply relying on straightforward web requests. Each episode publishes verbatim transcripts of the agent's activities, highlighting both its successes and failures. Additionally, 113 popular websites are evaluated weekly for their usability by AI agents, with public scoring reports detailing how well they conform to agent-friendly standards. This initiative is significant for the AI/ML community as it directly addresses the challenge of how effective current web infrastructures are for automated agents. By systematically scoring websites and providing transparency concerning the AI's performance, Agentability not only identifies barriers that AI agents face on the web but also encourages remedial actions for improving site compatibility. The empirical data generated could influence future developments in web design and AI integration, fostering a more accessible digital landscape tailored for autonomous agents.