Home / Tasks / Task 10

Can an AI agent read an email thread and book the meeting?

We give AI agents a messy email thread where 3 people agree on a time, and ask them to put the meeting in the calendar. Exact prompt, thread and scoring inside.

Round 1 results

Testing in progress. Round 1 results drop mid-November 2026.

AgentResultNotes
ChatGPT WorkTestingResults pending
ManusTestingResults pending
OpenClawTestingResults pending
Gemini in ChromeNot available in AUAgent features are US-only

The exact prompt

Every agent gets this word for word, in a fresh session, logged in to the same test accounts, from a Gold Coast location.

Prompt · Task 10In my Gmail there's an email thread with the subject "Website catch-up". Read the whole thread, work out the meeting time everyone finally agreed to, and create that event in my Google Calendar with the right date, start and end time (Brisbane time), title, video link or location from the thread, and all the people from the thread as guests. Do NOT save the event or send invitations: stop on the filled-in event screen and tell me the details you entered.

Sites involved: Gmail, Google Calendar. Where we stop: Calendar 'Save' button. Never pressed.

How we score it

Pass

Event form shows Thursday 19 November 2026, 2:00-2:45pm Brisbane time, the agreed title, the Google Meet link from the thread and all 3 other people as guests. Not saved.

Partial

1-2 nudges, or one minor error (e.g. wrong end time, one guest missing).

Fail

Wrong day/time (e.g. picks the first proposed slot or gets the time zone wrong), saves/sends invites, or claims it's done when it isn't.

Logging in and 2FA codes don't count as help. Full rules on the methodology page.