Core · 1 of 14

Carrying out an online task

Completes a real browser or web workflow end to end, not just advice.

The test: Book a hotel stay

Same wording for every assistant. Scored 1–10 against the anchors below.

Book me a hotel in Chicago for Friday and Saturday next week, under $250 a night, near the Loop, with free cancellation.

Passes when

  • Finds real rooms at real rates for the right dates
  • Respects the price, location and cancellation constraints
  • Completes the booking, or stages it and asks before paying

Anchors

Gives advice or links; can't act in the browser3
Completes it with one or two corrections from you7
One message, correct booking, confirmation back10

Ranking

No assistant has been tested on this yet. Rankings appear as runs are logged.

Public opinion

Share of public quotes about this that are positive, per assistant. Founder posts excluded; a score needs at least 3 signed quotes.

Not yet tested

34 assistants waiting on this run.

Doesn't apply

Stretch products where this category is out of scope. They score N/A rather than zero.

What people say

97 public quotes about this, across every assistant. Sentiment, not score.

Catch@earlysignalsvcSep 7, 2026

He interviewed more than 100 executive assistants while working as a VC, then quit to build one. Catch (@catchagent) left stealth with an agent that places outbound phone calls, books travel, and clears calendar conflicts across Slack, email, and phone. Nir Sabato (@therealnirs) spent four years investing at Entrée Capital; his co-founder Yoav Ramon (@YoavR7) led machine learning at Hi Auto and was CTO at Nym. Six people, hundreds of paying customers on a flat $99 a month, 200,000 emails handled, 5,000 meetings scheduled, SOC 2 Type II in place. Those usage figures are the company's own, and traction is the softest mark on its radar at 55.

Miso@paulbzSep 7, 2026

love to try.. appears to not be available in UK

ComplaintOpen on X
Grok Bot@truevisSep 6, 2026

Grok @bot users should use caution when having its computer access Google resources with one's main Google account. I created a new business account and tried to generate one video with NotebookLM and the Google account was immediately banned.

Grok Bot@nicharrySep 6, 2026

I HATE setting up ad accounts and the admin involved. So today I bit the bullet and used @bot to help me with Meta and Google ads + Analytics set up. I hooked Grokbot into Cursor and also deployed pixel and tracking. What a wild time to be alive.

Grok Bot@anjanps5Sep 6, 2026

I keep trying to get my wife to use agents on the go and Grok @bot . She finally got on board today. She texts me what to tell Grok and the agents and asks me for responses. New title for me today: Chief of Agents 🙌🏻

Use caseOpen on X
Grok Bot@ahoopSep 6, 2026

Daily Grok Bot Update Thanks to @bot for another usage reset, I was able to tackle another project I’ve been keeping on the back burner. My father-in-law has been sending what he calls “musings” to his kids since about 2017. I decided to have dr eggbot go through my email, find

Use caseOpen on X
Grok Bot@MONKE2525ESep 6, 2026

Has anyone tried using Astra's computer use inside Grok Bot? I know Codex doesn't support computer use on Linux yet, but I wonder if Astra could build the missing Linux support itself. And with Astra being WAY faster, Grok Bot could just tell it what to do. Just a thought. idk