For this test, I sent all AI agents a long paragraph explaining that my partner and I are planning to explore China late next year on a two-week adventure. I requested a first draft of a potential itinerary including: flight, hotel, nightlife and food recommendations. In addition, I asked the AI agents to trace a route combining historical moments and atmospheres of big cities. I can’t stand constantly changing locations, so I told the agents that I wanted to stay in up to four different hotels.
Notably, all chose identical routes across China, potentially ushering in a new era of the Lonely Planet effect, where groups of travelers are enticed to visit similar places, shops and restaurants.
Muse: 5/5. This draft itinerary was extremely easy to read, with visual charts and simple tips.
Fly from SFO the week after Golden Week, for a non-stop flight to Beijing, non-stop return from Shanghai. Four hotels, no going back: imperial history first, Sichuan cuisine and pandas in the middle, Shanghai nightlife lastly.
Points: 4/5. Dots’ draft was solid but excessive, running to 24 pages.
AI Agent Excerpt: My recommendation: fly to Beijing, take the high-speed trains to Xi’an and Chengdu, then fly to Shanghai and return home. You get imperial history, the Terracotta Army, Sichuan cuisine, and a big city finale without the night train or extra hotel move.
Bot Grok: 4/5. Grok Bot’s output was easy to read and full of realistic and useful tips and suggestions.
A first travel plan of 13 nights for two: Beijing, Xi’an, Chengdu and Shanghai, departure from SFO on October 13 and return on October 27, after Golden Week.
Gemini Spark: 3/5. This route was good, although it lacked readability or finer details than the top-ranked routes.
An itinerary for a beginner traveler combining imperial history, Sichuan gastronomy, vibrant nightlife and the comfort of a 4-hotel center.
Instinct: 3/5. This route was decent and helpful, although I preferred other agents’ layouts.
Beijing in, Shanghai home, with Xi’an and Chengdu in between: four bases, no turning back, history straight away and the food and bars get better with each stop.
Claude Duo: 3/5. Although he did not use any images or other visual elements in his production, Claude Duet’s route was solid.
Fly to Beijing on Saturday October 9, 2027, then Beijing → Xi’an → Chengdu → Shanghai, and return home from Shanghai on Saturday October 23. That’s 13 nights in four hotels, two high-speed trains and a short flight.
Insomnia control
I had trouble sleeping after all the AI agent chaos, although the Vietnamese coffee probably didn’t help. What happens when you give these tools a vast and limitless task? Sometimes when I need help the most, it’s when I don’t even know where to start.
If I had a human assistant, I would Never send them a message in the middle of the night, then maybe that’s a plus for these digital assistants. The two agents who really failed at this task, Gemini Spark and Claude Duet, responded to me without the gentleness or situational context that a real person might.
Bot Grok: 5/5. This release made sense to me, and if the storyline were real, I would probably follow through on it.
I can set a reminder for later in the morning to check in on your mom and see how her leg is doing, so it doesn’t slip through your mind just yet.
Points: 4/5. While Dots and Gemini Spark both gave me a similar task – say what you think, and we’ll get to it later – Dots’ phrasing seemed more encouraging and direct.
Send me the three things that take up space in your head; messy is good, and I’m going to turn them into a short checklist with a clear first step. Then you can put the phone down.
Instinct: 4/5. Unlike some of this agent’s other responses, this suggestion seemed reasonable.
Here’s one you can do from your bed: look through the Chinese draft and tell me the one thing you would change. I’ll have v2 ready when you wake up.
Muse: 4/5. Another strong response from Muse. This agent seems aggressively designed to improve every day and become more aligned with what would help the user.
I will check today’s calendar and your self-sent email reminders for any conflicts or can’t-miss items (the check I should have done properly yesterday) and only bring it up at a reasonable time unless something is urgent to begin with.
Gemini Spark: 2/5. OK, this is good advice for a user who complains of sleeping problems. At the same time, the comment about my empty morning calendar was annoying. Insomnia isn’t just about the contents of my Google Calendar.
The best quick task right now is a mental download to get everything going on in your head down on paper so your brain can let it go, especially since your calendar is completely clear today, with no morning commitments waiting for you.
Claude Duo: 2/5. One of Claude Duet’s weakest answers. Although the two are technically blurred, I’m not terribly worried about revisiting my lunch order and work plans in the middle of the night.
Let’s close the two details from yesterday. Your Thursday slot is delivered and your lunch basket has been covered by the DragonEats order. Say the word, and I’ll mark both as done, and then you can go back to sleep.
Best Overall: OpenAI Points
Yes, it’s still behind an expensive paywall, so I would suggest most people wait until OpenAI releases this agent for free, or at least with the $20 per month plan, before trying it. Despite this, Dots performed the best in these tests and failed in a task quite specific to my niche.
In the month since its release, Dots already seems more polished. I prefer a separate app to the current ChatGPT integration, although the smartphone widget option is a decent compromise. Points for victory.
The most accessible: Meta Muse
Muse is for normal people! And that’s very good. It performed well in testing, coming in a close second on the scoreboard, and was the easiest to set up. Oh, and it’s free. Despite Meta’s assurances that it responsibly handles user information in secure virtual machines, I still have concerns given the company’s past mishandling of user data.
Best interface: Anthropic Claude Duet
I tested this AI agent just a day after its official launch as a research preview, so some rough edges were expected. Although Claude Duet scored in the middle of the pack in most tests, its dedicated app was extremely user-friendly for navigating and tracking current tasks. I could see users obsessed with productivity and streamlined software experiences really clinging to Claude Duet over the next few months. Although you currently have to sign up for Anthropic’s waitlist to access Duet, and that’s behind a $20 per month paywall.
The most annoying: instinct
You might like the clean design that’s fully integrated with your messaging; I didn’t do it. While other WIRED writers praised Instinct, calling it so good that they didn’t care what happened to their private lives, the texting-only aspect struck me as a deal-breaker. I left the experience feeling too distant from what my agent was doing on my behalf and annoyed by his emoji reactions, like a grimace when I told the agent he missed my scheduling conflict.
Gn tech
