A 24-hour evaluation of six major AI personal agents on an iPhone tested proactivity, direction following, and browser control across real-world scenarios.
The tests covered tools from OpenAI, Meta, Google, Anthropic, and xAI, comparing performance, pricing models, and interface designs.
The test involved Meta's Muse, OpenAI's Dots, Instinct, Anthropic's Claude Duet, Google's Gemini Spark, and xAI's Grok Bot.
These apps vary significantly in how they work and their price. Meta's Muse and Instinct are both free, although Instinct works only via text messages and requires a special invitation code. OpenAI's Dots costs $100 per month and is built into ChatGPT, with the ability to use a special button on the phone screen. Google's Gemini Spark is a test feature within the Gemini app at a cost of $20 per month. xAI's Grok Bot costs $30 per month and runs in its own app, while Anthropic's Claude Duet costs $20 per month and was released in beta with a waiting list just before testing.
How were the tests conducted?
The assistants were assessed on their ability to take initiative, follow directions, and control the internet in five different scenarios. These included creating news updates, ordering meals, managing calendar conflicts, planning trips, and afternoon check-ins.
The test results
In the overall score, OpenAI’s Dots scored the highest, despite its high cost. Meta’s Muse was rated as the easiest and most affordable option for the average user during initial setup. Anthropic’s Claude Duet received positive reviews for its easy-to-use interface, while Instinct was rated as the most annoying assistant due to its text-only mode and emoji reactions.
READ ALSO
OpenAI introduced dots, a new digital assistant for subscribers
G
Hacks.gr on Google
Set it as a preferred source for cybersecurity updates.
Preferred source
end of article
Top comments (0)