A day with AI agents: testing the best tools for automation

In October 2026, a journalist tested six major AI agents—Meta’s Muse, OpenAI’s Dots, Instinct, Anthropic’s Claude Duet, Google’s Gemini Spark, and xAI’s Grok Bot—to evaluate their performance in automation tasks. The agents varied in pricing and features, with Muse being free and dominating app charts since September 2026, while Dots required a $100 monthly subscription. The tests focused on task execution, browser control, and proactive behavior, with Muse and Dots showing better adherence to instructions. One user reported an incident where an AI agent inadvertently shared financial data with colleagues, highlighting privacy concerns. The journalist also assessed the agents’ ability to compile tech news briefings, with Gemini Spark generating the most compelling pitch. Despite improvements, the tools remain nascent, with limitations in cross-app coordination and data integration. The tests included daily tasks like ordering lunch, where DragonEats and Freshroll were popular choices, though Gemini Spark failed to meet the user’s needs. The agents also struggled to detect calendar-email conflicts, with only two successfully identifying a scheduling clash. While the tools offer convenience, the journalist emphasized the need for better integration and user trust. The results suggest that while AI agents are improving, they still have significant room for growth and refinement.